• assigned source
• digital museum collection item (example)
• image
• sound recording (e.g., from radiooooo.com)
• dataset (e.g., list of constitutions (example), www.slavevoyages.org, orbis.stanford.edu)
• computer game
Your creation should demonstrate its clear, significant value, and have a professional appearance. Your creation should be accompanied by a commentary (300 to 500 words) explaining your goals for the project and how you sought to meet them (unless you are doing a proposal).
Your synthesis should demonstrate a clear, significant value, and have a professional appearance.
This table shows the distribution of grades in the course, slightly randomized to preserve privacy.
June 20, 2024
June 16, 2024
The history of racism and racist violence: monuments and museums, join the aha.
The AHA brings together historians from all specializations and all work contexts, embracing the breadth and variety of activity in history today.
This page contains words or ideas that might be offensive to modern readers. To maintain the accuracy of historical documentation, the content is reprinted in its entirety as it was originally published. This accurate reproduction of original historical texts therefore contains words and ideas that do not reflect the editorial decisions or views of the American Historical Association.
1 introduction, 4 discussion, author contributions, supplementary data, conflict of interest.
Isar Nassiri, Andrew J Kwok, Aneesha Bhandari, Katherine R Bull, Lucy C Garner, Paul Klenerman, Caleb Webber, Laura Parkkinen, Angela W Lee, Yanxia Wu, Benjamin Fairfax, Julian C Knight, David Buck, Paolo Piazza, Demultiplexing of single-cell RNA-sequencing data using interindividual variation in gene expression, Bioinformatics Advances , Volume 4, Issue 1, 2024, vbae085, https://doi.org/10.1093/bioadv/vbae085
Pooled designs for single-cell RNA sequencing, where many cells from distinct samples are processed jointly, offer increased throughput and reduced batch variation. This study describes expression-aware demultiplexing (EAD), a computational method that employs differential co-expression patterns between individuals to demultiplex pooled samples without any extra experimental steps.
We use synthetic sample pools and show that the top interindividual differentially co-expressed genes provide a distinct cluster of cells per individual, significantly enriching the regulation of metabolism. Our application of EAD to samples of six isogenic inbred mice demonstrated that controlling genetic and environmental effects can solve interindividual variations related to metabolic pathways. We utilized 30 samples from both sepsis and healthy individuals in six batches to assess the performance of classification approaches. The results indicate that combining genetic and EAD results can enhance the accuracy of assignments (Min. 0.94, Mean 0.98, Max. 1). The results were enhanced by an average of 1.4% when EAD and barcoding techniques were combined (Min. 1.25%, Median 1.33%, Max. 1.74%). Furthermore, we demonstrate that interindividual differential co-expression analysis within the same cell type can be used to identify cells from the same donor in different activation states. By analysing single-nuclei transcriptome profiles from the brain, we demonstrate that our method can be applied to nonimmune cells.
EAD workflow is available at https://isarnassiri.github.io/scDIV/ as an R package called scDIV (acronym for single-cell RNA-sequencing data demultiplexing using interindividual variations).
Although single-cell analyses are beginning to unravel the molecular aetiology of diseases, most studies incorporate average gene expression profiles across bulk tissues, which frequently mask variation between individuals visible at single-cell resolution ( van der Wijst et al. 2018 , Yazar et al. 2022 ). The variability between individuals remains an understudied aspect of co-expression relationships between molecules at the single-cell level. Establishing molecular co-dependencies may break down due to genetic variation, uncoupling of transcription and splicing, or it could be caused by pathological insult. The regulation of downstream markers may be disrupted by altered interindividual co-expression relationships, which in turn affect the expression pattern of entire biological pathways ( Johansen et al. 2023 ).
Previous studies have shown that SNPs can alter co-expression relationships on an individual donor basis by looking for evidence of allele-specific correlation ( Fairfax et al. 2012 , van der Wijst et al. 2018 , Oelen et al. 2022 ). Here, we develop a computational method, expression-aware demultiplexing (EAD), that harnesses variation of interindividual co-expression signatures at single-cell resolution to stratify donors and improve the demultiplexing of pooled scRNA-seq data ( Fig. 1 ).
Workflow for computational demultiplexing of unrelated individuals in scRNA-seq. (a) First, we demultiplex pooled samples (vireo) ( Huang et al. 2019 ) using genetic differences inferred from scRNA-seq data (cellsnp-lite) ( Huang and Huang 2021 ). (b) Next, for each pair of individuals in the output of genetic demultiplexing, we estimate accurate gene expression values for all genes. ( g ) per cell using the gene expression recovery for single-cell RNA sequencing (SAVER) ( Huang et al. 2018 ). (c) We apply the LASSO to obtain the most representative subset of genes ( s ) ( Nassiri and McCall 2018 ) related to query gene ( q ). (d) We apply differential gene correlation analysis between pairs of query gene ( q ) and related genes selected by LASSO ( SG ), to identify the top first DCE genes interindividual (see section 2). (e) The co-expression patterns of the top first DCE genes ( q and g ) are used to fit a mixture model and reconstruct the sample identity of each cell.
The process of separating cells from multiple samples pooled in a single batch is referred to as demultiplexing in this article. Multiplexing can increase the number of donors that can be tested, reduce experiment costs, address batch effects, and make large-scale sample operations feasible. Multiple approaches can be employed to demultiplex pooled single-cell gene expression profiles, including DNA oligonucleotide tagging and natural genetic variation ( Huang et al. 2019 ).
Inspired by the differential co-expression (DCE) analysis of two genes between biological groups, we introduce an optimized statistical method for donor stratification guided by variation of co-expression signatures at single-cell resolution ( McKenzie et al. 2016 ).
Given cells with known sample identity, we demonstrate that differences in gene–gene relationships exist between individuals by providing multiple examples. The results show top interindividual DCE genes provide a distinct cluster of cells per individual and display the enrichment of cellular macromolecular super-complexes, or organelles related to metabolism (e.g. mitochondria and ribosomes).
We applied this approach to samples from isogenic inbred mice and showed that by controlling genetic and environmental effects, we can resolve interindividual variations related to metabolic pathways.
Sequencing data for evaluation and application covers 124 10× single-cell RNA-seq samples in 14 batches and 30 bulk RNA samples.
During library preparation for the first and second datasets (see sections 1.1.1 and 1.1.2), every sample (a group of cells that originated from the same donor) was assigned a specific index sequence. Sample indices were incorporated into the sequencing primers on Illumina sequencers. After sequencing, by using the 10× mkfastq pipeline the names of sample index sets were automatically identified and reads per sample were merged into the FASTQ files. Supplementary Tables 4 and 5 included reference tables with the complete set of index codes per donor. Accuracy and reliability of the embedded sample index sequences are hallmarks of the 10× Genomics platform. This provides a solid foundation for evaluating the performance of demultiplexing algorithms without introducing errors from the ground truth itself.
Eight metastatic melanoma (MM) patients were included in the first scRNA-seq data set, which contains human peripheral blood mononuclear cells (PBMCs) with known donor labels ( Fairfax et al. 2020 ). In this dataset, sample indices were used to label cells per sequencing sample without errors. By using sample/donor indices, we were able to map demultiplexed cells using the expression-aware approach to their original sample donors and assess their performance in singlet assignment ( Supplementary Table 1 ). Two sample pools cover 16 samples pre- and posttreatment at day 21 (eight pre- and eight posttreatment) with immune checkpoint blockade therapies including Nivolumab (NIVO) + ipilimumab (IPI) or Pembrolizumab (Pembro) ( Fairfax et al. 2020 ). We chose this dataset to show that interindividual differences and no other confounding factors, such as cell type, underlie differential gene co-expression across individuals. We used the Cell Ranger pipeline (v7.0.1), the GRCh38 reference genome, and 5′ R2-only chemistry to process the data set. The analysis detected up to 36 842 cells per pool, 56 850 mean reads per cell, and 1768 median genes per cell. Raw data for single-cell sequencing datasets have been deposited on the European Genome–phenome Archive ( https://ega-archive.org/studies/EGAS00001004081 ).
The second scRNAseq data set consists of a pooled sample with known donor labels from six isogenic mice. The samples were exposed to a topical TLR7 agonist, Imiquimod, to induce a systemic lupus erythematosus-like phenotype or vehicle control. In this dataset, cells were labelled per sequencing sample using sample indices ( Supplementary Table 3 ). We applied this dataset to show that by controlling genetic and environmental effects, we can resolve interindividual variations related to metabolic pathways. The Cell Ranger pipeline, mm10-2020-A reference genome, and single-cell 3′ v3 chemistry were employed for data processing. The analysis detected up to 12 685 cells per pool, with an average of 40 101 mean reads per cell and 1553 median genes per cell.
The third data include five single-cell multi-omics (RNA-seq + ATAC-seq) batches with unknown donor labels. The dataset consists of circulating haematopoietic progenitor cells samples from seven healthy controls, 15 sepsis patients, and eight convalescent samples (six samples per batch) ( Kwok et al. 2023 ). The genetic demultiplexing pool samples were based on the use of an extra 30 bulk RNA-sequencing profiles from the same individuals. This dataset was chosen to evaluate the utility of EAD for challenging biological models involving a mixture of heterogeneous cell types and donors. The data set was processed using the Cell Ranger pipeline, GRCh38 reference genome, and Multiome chemistry. The analysis detected up to 14 806 cells per pool, 49 892 mean reads per cell, and 3250 median genes per cell. Raw data for single-cell sequencing datasets have been deposited on the European Genome–phenome Archive ( https://ega-archive.org/studies/EGAS00001006283 ) ( Kwok et al. 2023 ).
The fourth dataset includes a scRNA-seq of human MAIT cells ( Garner et al. 2023 ). We used a subset of the original dataset, comprising five channels of a Chromium Next GEM Chip K. MAIT cells from three donors were either left unstimulated or activated with a TCR, cytokine, or dual TCR + cytokine stimulus. Cells from each donor-condition combination were labelled with TotalSeq-C hashtag antibodies (12 total), pooled, and split across the eight channels of the Chromium Chip ( Garner et al. 2023 ). The Cell Ranger (v7.0.1) count pipeline was used to process FASTQ files for gene expression. Next, hashtag oligo demultiplexing (HTO) and the extraction of singlets, doublets, and negative cells from multiplexing experiments were performed using the HTODemux algorithm ( Stoeckius et al. 2018 ) implemented in the Seurat tool ( Hao et al. 2024 ). We chose to utilize this dataset to demonstrate how interindividual DCE patterns can enhance cell hashing demultiplexing outcomes. The analysis detected up to 16 251 cells per pool, 53 413 mean reads per cell, and 3525 median genes per cell. Raw data for single-cell sequencing datasets have been deposited on the Gene Expression Omnibus (GEO) database ( https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE194187 and https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE194189 ) ( Garner et al. 2023 ).
The fifth dataset includes ventral and dorsal tiers of the substantia nigra (SN) and the cortex (middle frontal gyrus) from five healthy donors ( Agarwal et al. 2020 ). The total number of samples is 12, which includes two replicate samples of SN. The sample libraries contain genes that vary by nuclei from 607 to 3364, and mean Reads that range from 18 377 to 59 513 in both regions ( Agarwal et al. 2020 ). To obtain processed single-nuclei RNA-sequencing matrices, use the accession code https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE140231 from the Gene Expression Omnibus. This dataset was used to show that our method can be applied to nonimmune cells.
Sequencing data was processed by Cell Ranger (v7.0) pipelines to create a feature barcoding and gene expression library. The filtered gene–cell matrix generated by the cell ranger was converted from an HDF5 gene–cell matrix to a gene–cell count matrix using the cell ranger mat2csv command provided by 10× genomics. We applied the scater package to filter out single-cell profiles that were outliers for any metrics, as they are considered low-quality libraries ( McCarthy et al. 2017 ). The SCRAN package was applied to detect and remove doublets using expression profiles as described in Dahlin et al. (2018) . Cell type annotation on scRNA-seq data was performed using the scQCEA tool ( Nassiri et al. 2023 ).
Single-cell RNA-sequencing studies only sequence a small portion of the transcripts present in each cell. This leads to inaccurate quantification of genes with low or moderate expression levels. We used expression recovery methods to provide accurate expression estimates for all genes in order to address this challenge ( Huang et al. 2018 ). The SAVER method was our preference because it recovers the relationship between two marker genes, which we are aware do not correlate ( Huang et al. 2018 ).
We introduce an optimized statistical method to detect DCE patterns using single-cell data from two individuals ( McKenzie et al. 2016 ).
We use cellsnp-lite (v1.2.2) ( Huang and Huang 2021 ) and vireo (vireoSNP/0.3.2) ( Huang et al. 2019 ) to infer genetic variants from scRNA-seq data in the first step. Gene–cell count matrices for all possible pairs of individuals are generated using vireo’s best-proposed donor for each cell (or alternative resources, including the use of HTO). The SAVER ( Huang et al. 2018 ) tool is used to transform the gene–cell count matrix per pair of individuals (see section 1.2). The glmnet R package ( Engebretsen and Bohlin 2019 ) is used to select representative gene subsets in the expression profile for each pair of individuals. The calculation of correlation coefficients ( r ) for all possible pairs of selected genes is done using lasso. Interindividual differential gene correlation analysis (IDCA) was conducted for two donors ( D 1 and D 2 ) and genes ( G 1 and G 2 ) using correlation coefficients as described below.
We use a bivariate mixture model for uncovering correlation classes (donors) for each pair of differentially correlated genes. We reconstruct the identity of each cell by fitting a mixture model. The bivariate Gaussian mixture model is used to measure the joint variability between the expression profile of an indicated cell and two donors (donor label proposed by genetic-based demultiplexing) using maximum posterior probability. We applied flexmix R package( Grun and Leisch 2008 ) to perform mixtures of regression models using the Expectation Maximization algorithm and model-based clustering.
In order to reconstruct the sample identity of each cell, a mixture model is fitted using the co-expression patterns of the top DCE genes. A mixture model is utilized to predict a cluster of cells that are not labelled and to reconstruct their sample identity. We consider all possible donor pairs for an indicated cell. This procedure is repeated for the top 10 genes that are differentially co-expressed. We verify that the cell is from a donor if we are able to assign it to the donor most of the time (number of assignments to an indicated donor equal or greater than the total number of pairs of donors minus 1). Reconstructing the sample identification of cells can be done using more liberal thresholds, such as the total number of pairs of donors minus 2.
Bulk RNA-seq profiles of 30 samples were used for genetic variant calling using GATK (Genome Analysis Toolkit) ( Deelen et al. 2015 ). Our first step was to align the bulk RNA-seq reads to the hg38 reference genome using the Burrows-Wheeler Alignment tool ( Li and Durbin 2009 ). Next, we used GATK’s HaplotypeCaller tool to find genetic variants in the aligned reads ( Van der Auwera et al. 2013 ). Using BCFtools, we removed false positives and low-quality variants from the variants after calling them ( Li 2011 ). Bulk RNA-Seq genotypes were inputted into the demuxlet tool with a posterior probability (PRB = 1) of singlet assignment ( Kang et al. 2018 ).
The first step in assigning genetic donors to samples is to perform SNP genotyping using CellSNP-Lite (v1.2.3) ( Huang and Huang 2021 ) in a given data set. We followed the cellSNP-lite manual’s recommended default parameters. Using the cell data from cellSNP-Lite as input, we demultiplexed using vireoSNP (v0.5.8) ( Huang et al. 2019 ).
The pathway enrichment analysis ( Wu et al. 2021 ) of interindividual DCE genes ( McKenzie et al. 2016 ) was performed using R packages. Box plots and dot plots were generated using ggpubr (v0.2) and customizing ggplot2 ( Almeida et al. 2018 ).
EAD workflow is available at https://isarnassiri.github.io/scDIV/ as an R package called scDIV (acronym for Single-Cell RNA-sequencing data Demultiplexing using Interindividual Variations). Our implementation with variable selection and proper data structures has made the EAD computationally efficient and can be run on a laptop with 16 Gb of memory and two 3.5-GHz CPUs. To run the tool on multiple servers simultaneously for large datasets, users should use a shell script. The package website has documentation that includes examples.
We developed a generic five-step workflow for demultiplexing scRNA-seq data using interindividual variation in gene expression ( Fig. 1 ). First, we infer genetic variants from scRNA-seq data ( Huang and Huang 2021 ) and demultiplex pooled samples ( Fig. 1a ). Gene–cell count matrices are generated for all possible pairs of individuals by utilizing the best-proposed donor for each cell in the previous step. The gene–cell count matrix is transformed by an expression recovery method per pair of individuals to provide precise gene expression values for all genes per cell ( Fig. 1b ) ( Huang et al. 2018 ). We apply lasso (least absolute shrinkage and selection operator) to find compact and representative gene subsets in the expression profiles to improve the accuracy and reduce the redundant downstream number of computations steps ( Fig. 1c ) ( Nassiri and McCall 2018 , Yang et al. 2021 ). Next, we performed an individual-specific co-expression analysis searching for altered co-expression patterns of gene pairs interindividual ( Fig. 1d ) (see section 2). A mixture distribution of correlation classes for each pair of differentially correlated genes is used to fit a mixture model. We reconstruct the identity of each cell based on the similarity of their gene expression with donor-specific clusters using the mixture model ( Fig. 1e ). Cells are assigned if expression-aware and genetic-based demultiplexing propose the same best singlet ( Fig. 2 and Supplementary Fig. 1 ). In this way, the combined method can achieve greater accuracy than the SNP-based or barcode-based methods by selecting the element-wise maximum of the two demultiplexing results.
An example of EAD is to assign an indicated cell to one of seven individuals using a mixture model. Genetic-based demultiplexing already suggested that the cell most likely belongs to donor-0 using the partial genotypic data from state of the individuals in a pool of donors. Now, using the gene expression profiles, we want to check the best-guess assignment obtained from genetic-based demultiplexing. (a) We select the top first pair of DCE genes per donor-0 compared to other donors. The expression pattern of a pair of DCE genes is used to create distinct clusters of cells across individuals. (b) We use a mixture model to predict a cluster of unlabelled cells and reconstruct their sample identity. (c) For an indicated cell (TACTCATCAGCTTCGG-1), we consider all possibilities and most of the time the cell is assigned to donor 0. We confirm that the cell belongs to donor 0 if we successfully assign it to donor 0 for an equal or greater number of pairs of donors, minus 1.
There are multiple methods that can be used to label cells or nuclei with antibodies-based oligonucleotides ( Stoeckius et al. 2018 , Gaublomme et al. 2019 ). By sequencing the cells’ RNA molecules with HTOs, a matrix of count for HTOs per droplet is produced and utilized for demultiplexing of the sample pools. Barcode-based multiplexing methods may be used to combine cells from the same donor and different stimulation conditions into sample pools ( Garner et al. 2023 ). In these circumstances, the genetic-based demultiplexing cannot differentiate between two samples that came from the same donor but had different stimulation conditions. Our EAD workflow uses alternative methods like HTODemux ( Stoeckius et al. 2018 ) to estimate gene expression for each pair of individuals ( Fig. 1b ) in these cases. In order to evaluate EAD, we utilize various datasets that employ diverse methods to label cells in scRNA-seq data for each sequencing donor. These approaches involve sample indices, genetic variation derived from bulk RNA-sequencing profiles of the same samples, and HTOs.
Our initial assessment of EAD was based on synthetic sample pools from eight MM patients ( Fairfax et al. 2020 ). To perform a preliminary evaluation of EAD, we mixed intact donor cells, and refer to them as the ‘synthetic sample pool’. A unique identification number was given to every cell in the synthetic sample pool to reveal its true origin. By using mixed cells with known donor identities, we were able to evaluate the performance of demultiplexing algorithms without introducing errors from the ground truth itself.
Peripheral blood samples were obtained from patients with MM who were treated with immune checkpoint blockade ( Fairfax et al. 2020 ). Samples were collected both before and after the initial treatment cycle. PBMCs were used to isolate monocyte and T cells. Monocyte and T cells were mixed in suspension and the Chromium 10× system was utilized to process the single-cell transcriptome. During library preparation for scRNA sequencing, libraries were tagged using unique indices per donor ( Supplementary Table 1 ). Each sequencing run involved pooling (multiplexing) multiple libraries and sequencing them together ( Fairfax et al. 2020 ). The sequencing run concluded with demultiplexing, and the reads produced were separated into different FASTQ files according to donor indices.
Sorting a pile of laundry (sequencing reads) with a tag (donor index) that identifies the owner (donor) is an analogy to demultiplexing in sequencing. This is not in line with the definition of single-cell sample demultiplexing in this article. In this article, demultiplexing is similar to sorting a pile of laundry (sequencing reads) without tags for each item, by incorporating other indicators like size or personal clothing style (genetic variation) ( Howitt et al. 2023 ).
We demultiplex monocyte and T cells separately in each synthetic sample pool ( Fig. 3a and b ). In the original paper, subsetting was performed to select T cells expressing CD8A, CD8B, and CD3D, and monocytes expressing CD14. We applied further subsetting to eliminate heterogeneous cell populations, including monocytes expressing CD3D, CD3E, CD3G, CD8E, or CD19, and T cells expressing CD14 or CD19.
(a) The results of quality control and the number of called variants for two single-cell sample pools made of 16 samples. (b) Example of top DCE genes (EEF1A and RPS4X) in PBMCs, distinguishes clusters of cells per individual. Pairwise correlations can be visualized as a network. The result shows that EEF1A and RPS4X display co-expression only in donor 1 (red arrow), which could not have been detected based on all individual cells or donor 2. (c) The results of ontology gene set enrichment analysis show a significant association of interindividual DCE genes with the regulation of metabolic processes. The enrichment analysis of cellular components in the dot plot shows associations with mitochondria, ribosomes, cellular macromolecular super complexes, or organelles related to metabolism. (d) Comparison between the methods in terms of prevalence (abbreviations: EAD, expression-aware demultiplexing; GD, genetic-based demultiplexing; mixed, a combination of ED and GD results). According to prevalence, a combination (mixed) of GD and EAD leads to better results for Monocytes and T cells. The colour density reflects the continuous data range to compare values. Lower values are shown to be the most profitable. The range of prevalence values is between 0 and 1.0.
We compared the performance of overall demultiplexing matrices generated based on genetic-based (GD), EAD, and mixed methods for demultiplexing results against that from the known sample labels ( Fig. 3d and Supplementary Table 2 ). We accept the best singlet proposed by GD for unassigned cells if it is consistent with the donor proposed by EAD and call it mixed demultiplexing ( Huang et al. 2019 ).
To calculate the accuracy and evaluate the results, we define true positive (TP) as a donor assignment result that correctly assigns the donor according to the known label. The term false positive refers to a donor assignment result that incorrectly assigns the donor based on the known label. True negative (TN) is characterized by an unassigned donor where the best guess incorrectly suggests a particular donor based on the known label. A false negative is defined as an unassigned donor result that correctly suggests a particular donor based on the known label.
The results showed that the combination of GD and EAD results improves the prevalence and harmonic mean of precision and sensitivity (F1 score) ( Fig. 3d and Supplementary Table 2 ). The accuracy of the two approaches is comparable, and mixed results improve accuracy ( Supplementary Table 2 ).
Enrichment analysis revealed that genes that are differentially co-expressed between individuals are strongly linked to the regulation of metabolic processes and represent metabolic differences between individuals ( Fig. 3c ). Therefore, our method to detect interindividual variation in gene expression could be applied to provide insight into challenges presented by interindividual differences in the responses to nutrition and obesity, cardiovascular and endocrine research, and comprehending the alterations that occur with age and the ensuing neurodegenerative conditions ( Manach et al. 2017 , Lotta et al. 2021 , Johansen et al. 2023 ). Figure 3b shows an example of top DCE genes ( EEF1A and RPS4X ) associated with memory decline in normal aging and Alzheimer’s disease, which distinguishes clusters of cells per individual ( Beckelman et al. 2016a , b ). Only donor 1 showed co-expression of EEF1A and RPS4X , which was not present in all cells or donor 2. Changes in interindividual co-expression relationships could have an impact on the regulation of downstream markers whose expression pattern influences entire biological pathways.
We found that 939 cells were not assigned to any donor. The location of detected unassigned cells on the knee plot showed aggregation after the inflection point ( Fig. 4a and b ). In addition, unassigned cells were mostly found in the bottom-left corner of the quantification plot, as shown in Fig. 4 . It means filtering out nonrelevant cells can improve the accuracy of demultiplexing, especially for samples with a low number of cells. In general, the UMAP projection and number of genes per cell plots showed that unassigned cells in EAD tend to be assigned to regions with low number of genes per cell (Wilcoxon test P < .001) ( Fig. 4 ).
(a, b) The knee, quantification, and UMAP plots show the location of unassigned cells in two sample pools. We ensure that the background red-coloured cells appear on top by dividing the points into different layers and plotting the red points after the black points ( Nassiri et al. 2023 ). Number of detected gene plots shows an association between cell assignment and the number of genes per cell. (c) Boxplots show a significant difference in the mean number of genes per cell across the classes of assigned and unassigned cells using EAD.
We were able to control confounding factors by removing single-cell profiles that did not fit any metric and in vitro separating T cells and monocyte cells. This allowed us to show that patterns of differential gene co-expression across individuals are the consequence of interindividual gene expression differences and no other factors such as cell type ( Fig. 3b ).
Synthetic sample pools are valuable for evaluating scRNA-seq demultiplexing approaches, but they have some shortcomings to consider. The complexity of real pooled single-cell RNA-seq data systems may not be fully captured by synthetic sample pools. Furthermore, a synthetic sample pool may not cover the full range of technical variations that can happen during actual sample pooling. If the synthetic sample pool does not fully replicate real-world scenarios, the observed patterns could reflect the limitations of the synthetic model.
We further evaluate the effectiveness of EAD using several heterogeneous cell populations in the following sections.
In this model, both treatment group and sex might be expected to alter the immune transcriptome. Therefore, we stratified individuals based on the type of treatment and sex as confounding variables to consider their influence on DCE patterns ( Bongen et al. 2019 ). We found no consensus interindividual DCE patterns for matched pairs based on sex (e.g. Male and Male) or type of treatment (e.g. Pembro and Pembro) ( Fig. 5a ).
(a) Confounding variables influence the variation in gene expression between individuals. We stratified melanoma patients based on the type of treatment with immune checkpoint blockades (Ipilimumab + Nivolumab (Ipi + Nivo) or Pembrolizumab (Pembro)) and sex (male or female). If these factors cause DCE patterns among individuals, we expect the appearance and disappearance of the top first examples of DCE genes across classes. We found no such an accumulation. For example, the expression of EEF1A1 and RPS4X represent individual 7 and we do not see a similar DCE pattern for matched pairs of individuals based on treatment or sex (e.g. Male and Female). (b) The results of quality control and the number of called variants per cell for a single-cell pool sample made of 6 isogenic mice. We applied the scater package to filter out single-cell profiles that were outliers for any metrics, as they are considered low-quality libraries ( McCarthy et al. 2017 ). (c) A pool of six samples with known cell labels from isogenic individuals was used as input for an EAD workflow. DCE patterns across pairs of donors could not distinguish interindividual differences in the gene expression including genes related to the metabolic pathway (e.g. ugp2 and Miox genes), and we only see differences related to the treatment (e.g. sample 4 versus sample 2).
We evaluated the effectiveness of the expression-aware approach in demultiplexing cells from isogenic individuals. We pooled cells with a known label from six isogenic mice exposed to a topical TLR7 agonist Imiquimod to induce a systemic lupus erythematosus-like phenotype or vehicle control and tried to perform demultiplexing using the expression-aware approach ( Fig. 5b and c ). Sample indices were used to label cells per sequencing sample/donor in this dataset ( Supplementary Table 3 ).
Interindividual differences were not discernible through DCE patterns across classes of isogenic samples ( Fig. 5c ). The results indicate that it is challenging to identify variations in metabolic pathways for sample pools with very little genetic diversity among donors. Therefore, demonstrating the efficacy of expression-aware approach for demultiplexing these pooled samples was not possible.
The assumption that genetic and environmental differences lead to DCE patterns between individuals is supported by the absence of individual-specific signatures in isogenic laboratory mice ( Figs 3 and 5c ). These variables can be controlled to resolve interindividual differences in gene co-expression related to metabolic pathways ( Fig. 5c ). In conclusion, nonisogenic individual samples are suitable to be pooled for demultiplexing scRNA-seq data using interindividual variations in genetic and gene expression.
We considered more challenging scenarios involving five pooled sample (batch) and 30 donors (six donors per batch) to test if the method would work well with more heterogeneous cell populations ( Fig. 6a and b ) ( Kwok et al. 2023 ). From every donor, the nuclei of CD34+ circulating hematopoietic stem and progenitor cells (HSPCs) were extracted. Six samples with equal numbers of cells were combined to form a single pool. Each pool was then subjected to cell lysis and nuclear extraction. RNA libraries were sequenced after loading each pool of nuclei across four channels of the 10× genomics lane on the chip ( Kwok et al. 2023 ). In this context, a 10× lane would represent a batch loaded onto the chip. To differentiate this dataset from other datasets, we use batch and 10× lane interchangeably.
Demultiplexing with genotype reference (Demuxlet) provides a source of ground truth for benchmarking the performance of demultiplexing algorithms. (a and b) The results of quality control including the percentage of filter-out cells as a doublet and the number of called variants per cell for five single-cell pool samples (10× lanes) made of 23 sepsis and seven healthy individuals. (c) Percentage of correct (TP and TN), and incorrect assigned cells (FP and FN) using EAD, GD demultiplexing (vireo), and the combination of genetic-based and expression-aware approaches (Mixed). The outcomes are given for each 10× lane and for all 10× lanes. (d) The confusion matrices were utilized to generate the key metrics and give a comprehensive assessment of demultiplexing methods that do not require a reference genome. Abbreviations: EAD, expression-aware demultiplexing; GD, genetic-based demultiplexing; Mixed-D, mixed demultiplexing; NPV, negative predictive value; CSI, critical success index.
10× lanes in the scRNA-seq experiment of HSPCs included samples from healthy donors, sepsis donors, and donors with both sepsis symptoms and COVID-19 infection ( Kwok et al. 2023 ).
There are four methods that can demultiplex pooled single-cell RNA-seq without relying on reference SNP genotypes: vireo ( Huang et al. 2019 ), scSplit ( Xu et al. 2019 ), Freemuxlet ( Kang et al. 2018 ), and Souporcell ( Heaton et al. 2020 ). According to available benchmark studies, vireo outperforms other tools when it comes to demultiplexing pooled single-cell RNA-seq data without genotype reference ( Neavin et al. 2024 , Cardiello et al. 2023 ). Furthermore, vireo outputs have a high level of consistency with the outcomes of other tools ( Neavin et al. 2024 , Cardiello et al. 2023 ). Therefore, we opted for vireo as the genotype-free demultiplexer for pooled single-cell RNA-seq.
We utilized the Demuxlet tool and reference genotypes derived from bulk RNA-sequencing profiles of the same samples to determine the donor identity of every singlet ( Kang et al. 2018 ). SNPs in each individual’s genome were used in demultiplexing with genotype reference (Demuxlet) to identify individual donor identities per cell with high accuracy. This allows for a precise assessment of how other algorithms compare to this ground truth. The performance of EAD and GD (GD—vireo) ( Huang et al. 2019 ) methods for demultiplexing, which do not require a reference SNP genotypes, was evaluated in the next step using Demuxlet results.
The TP rate is determined by the fraction of cells with the corresponding genetic assignment (Demuxlet) for each possible EAD, Mixed-D or GD assignment. The fraction of cells with EAD, Mixed-D or GD assignments that are different from their genetic assignments is called the false positive rate (FP). The false negative rate (FN) is the proportion of cells that have genetic assignment and EAD, Mixed-D or GD un-assignment, and the best guess of EAD, Mixed-D or GD accurately indicates a specific donor. The TN rate is the proportion of cells that have genetic un-assignment and EAD, Mixed-D or GD un-assignment.
We used the sensitivity and F-score to evaluate performance as our main metric. The proportion of positive cases that are correctly identified by the evaluation method is represented by sensitivity. The F-score refers to the harmonic mean of precision and recall, which can range from zero to one. A higher F-score indicates superior performance.
Six samples from genetically distinct donors were represented by cells in each of the five 10× lanes in the data set. The data from each 10× lane was subjected to demultiplexing methods. A cell could be assigned to singlet, a singlet that corresponds to one of the six unique samples or unassigned by each demultiplexing method. Misclassifying true singlets as doublets can be a more significant error source in downstream analysis than misclassifying true doublets ( Wolock et al. 2019 , Howitt et al. 2023 ). Therefore, for vireo, we treat the doublets as unassigned and do not attempt to reassign them through mixed demultiplexing.
We next evaluated the donor assignments made by expression-aware, vireo, and Mixed-D methods against the genetic assignments from Demuxlet that were considered to be ‘ground truth’. Figure 6 exhibits mosaic plots of the confusion matrix for demultiplexing methods, divided by lanes. In all 10× lanes, Mixed-D had an increased TP and FP rate, but a decreased FN rate. Mixed-D and vireo demonstrated a consistent correct assignment (TP+TN) in all batches ( Fig. 6c ).
To quantitatively evaluate the performance of demultiplexing methods, we calculated the key metrics derived from the confusion matrix ( Fig. 6d ). Mixed-D had the most significant mean sensitivity, prevalence, and negative predictive value among all methods for every 10× lane. All 10× lanes displayed consistent F-scores and accuracy for Mixed-D and vireo ( Fig. 6d ).
While expression-based methods could not significantly improve the percentage of correct assignment (TP+TN) in 10× lanes of this heterogeneous dataset, Mixed-D could provide more confirmation of the vireo results and improve the accuracy of assignments ( Fig. 6c ).
Despite the diversity of cells, the top interindividual DCE genes produced a distinct cluster of cells per individual ( Fig. 7a ). Consistent with previous results, ontology gene set enrichment analysis for DCE genes showed a significant association with the regulation of metabolic processes ( Supplementary Fig. 2 ). While the top interindividual DCE genes are related to metabolic functions or organelles, the breakdown of co-expression patterns can represent context-specific pathognomonic signatures. An illustrative example occurs in the differential co-expression of genes specific to sepsis in the context of COVID-19 infection such as NEDD9 ( Rizzo and Yuan 2022 ), BACE2 ( Tang et al. 2022 ), and DHX30 ( Apostolidou et al. 2021 ). The application of interindividual DCE genes allows for the differentiation of subtle pathology-induced patterns in a context-specific manner.
(a) Example of the top interindividual DCE genes provide a distinct cluster of cells per individual. (b) There are no significant differences in chromatin accessibility across states observed in the case of the top DCE gene. (c) Association between cell assignment and the number of genes per cell. Some unassigned cells show a low number of genes per cell, which means filtering out unassigned cells can improve the accuracy of cell calls.
We investigated the potential for identifying interindividual single-cell expression variability using multiome ATAC. The results showed that chromatin accessibility within the same cell does not provide a similar explicit model for co-expression relationships between molecules at single-cell level ( Fig. 7b ).
The tSNE projection plot revealed that unassigned cells were accumulating in regions with low numbers of genes per cell ( Fig. 7c ). Therefore, the accuracy of cell identification can be improved by removing cells which have very few genes per cell. Functional enrichment can be combined with unassigned cells to maintain statistical power for pooled scRNA-seq data analysis as an alternative solution ( Fairfax et al. 2020 ).
The authors in the original paper ( Kwok et al. 2023 ) used the combined gene expression profiles to group cells according to their gene expression and chromatin data (seven clusters labelled as C1–7). The differences in gene expression between the clusters of the dataset are likely driven by their skew towards different lineages (e.g. C4: lymphocyte progenitors, C5: emergency granulopoiesis, and C7: normal granulopoiesis). We did not observe strong correlation between the proportion of cells in each cluster and number of unassigned cells including singlets and doublets.
These results demonstrate the potential utility of demultiplexing pooled single-cell RNA-sequencing samples using interindividual variation in gene expression in various biological models.
Multiple methods are available for labelling cells or nuclei with oligonucleotide-barcoded antibodies ( Stoeckius et al. 2018 , Gaublomme et al. 2019 ). A matrix of counts per HTOs per droplet is created by sequencing the cells’ RNA molecules with HTOs.
Using a scRNA-seq dataset of stimulated human MAIT cells ( Garner et al. 2023 ), we present a demonstration of how expression-aware and barcode-based demultiplexing methods can complement each other ( Fig. 8 ). We used a subset of the original dataset, comprising five channels of a Chromium Next GEM Chip K. MAIT cells from three donors were either left unstimulated or activated with a TCR, cytokine, or dual TCR+cytokine stimulus. Cells from each donor-condition combination were labelled with TotalSeq-C hashtag antibodies (12 total), pooled, and split across the eight channels of the Chromium Chip ( Garner et al. 2023 ).
(a) A distinct cluster of cells from the same donor but a different stimulation condition is present in an example of the top DCE genes. (b) Summaries of cell hashing demultiplexing results showing the number of singlets called per 10× lane along with the percentage of doublets and negative cells (both filtered out). (c–e) Evaluating the impact of different thresholds for EA sample demultiplexing on accuracy, percentage of assigned cells, and percentage of genetic-based demultiplexing doublets in EAD assignments by specifying various thresholds ( T ).
The dataset contained three donors and four conditions per donor per 10× lane. Genetic-based demultiplexing (vireo) was unable to differentiate two samples from the same donor but different stimulation conditions ( Howitt et al. 2023 ). Therefore, we were unable to produce results for mixed demultiplexing generated using genetic-based and expression-aware methods on this dataset. To estimate gene expression for each pair of samples ( Fig. 1a ) in our EAD workflow, we utilized the output of Seurat HTODemux ( Stoeckius et al. 2018 ) instead of vireo. We chose HTODemux as our preferred tool because it is one of the top three most effective methods and performs fairly well on different types of datasets with minor differences between its output and other tools ( Howitt et al. 2023 ). The EAD technique was capable of separating all 12 samples per 10× lane, including those that were derived from the same donor but different stimulation conditions ( Fig. 8a ). Several methods were used for demultiplexing of cell hashing data: BFFcluster ( Boggy et al. 2022 ), BFFraw ( Boggy et al. 2022 ), GMM-Demux ( Xin et al. 2020 ), MULTI-seq ( McGinnis et al. 2019 ), HTODemux ( Stoeckius et al. 2018 ), demuxEM ( Gaublomme et al. 2019 ), and DropletUtils hashedDrops ( Griffiths et al. 2018 , Boggy et al. 2022 ). Each algorithm assigns cells as singlets (one hashtag antibody detected), doublets (two or more hashtag antibodies detected), or negative (no hashtag antibody detected) ( Fig. 8b ). We treat the doublets and negatives as unassigned and do not attempt to reassign them through EA demultiplexing. To assess the performance of the EAD method, we compared its assignments to singlet calls from the six HTO demultiplexing tools. To calculate accuracy, we compared EAD classifications to HTO algorithm classifications, and then divided the number of concordant classifications by the total number of classifications.
By using 70 000 singlets from five 10× lanes, EAD was able to assign the 90% (62 682/70 000) of cells with an average of 93.2% accuracy ( Fig. 8c ). EAD’s results were similar across all HTO tools, but there was a slight decline for DropletUtils and BFFcluster ( Fig. 8c ). BFFcluster and DropletUtils determine doublets by thresholding barcode counts, resulting in the algorithm’s performance being highly dependent on the correct selection of this parameter ( Boggy et al. 2022 ). The default parameters recommended by cellhashR were used in this study ( Boggy et al. 2022 ). Adjusting the parameters could potentially yield more consistent results. It is probable that there are problematic singlets for the expression-aware method as it did not consistently assign 6.8% of singlets. Our next step was to delve deeper into the cases where demultiplexing with hashtag oligos and EAD is not in accordance.
Figures 1 and 2c explain that EAD takes into account the assignments for an indicated cell across all possible donor pairs. A cell that is successfully assigned to a donor for an equal or greater number of donor pairs minus 1 (0.9 of all the donor pairs that are possible), is confirmed as belonging to the donor by EAD ( Fig. 2c ). It is possible to utilize either more liberal or rigid thresholds ( T ) for EA sample demultiplexing to evaluate its impact on accuracy ( Fig. 8c–e ). According to the results, EAD produces balanced output for accuracy and the percentage of assigned cells when T is ≥0.8. Increasing the threshold to 1 could lead to an increase in accuracy to 99.9%, but at the expense of decreasing the percentage of assigned cells to 46% on average ( Fig. 8d ). Since there are three genetically distinct donors per 10× lane, we can use SNP-based multiplexing to estimate cells with mismatched SNP profiles (doublets). According to the findings, elevating the threshold for EA sample demultiplexing consistently decreases the number of detected doublets by SNP-based demultiplexing (vireo) ( Fig. 8e ). This trend remains the same even with different thresholds that yield similar percentages of assigned cells (e.g. 0.1–0.6) ( Fig. 8d and e ). By using the EAD with a high threshold for sample demultiplexing (e.g. T = 1), it is possible to obtain reliable results for cells with sufficient barcode counts for accurate classification, as evidenced by this result ( Fig. 8c and d ).
The 1968 discordant assignments occurred totally among singlet calls from six HTO demultiplexing tools implemented in cellhashR ( Boggy et al. 2022 ). cellhashR defines discordant cells when there is at least one hashing algorithm that produces a different result. The majority of discordant results are caused by the algorithms having slightly different thresholds, as demonstrated in the previous study ( Boggy et al. 2022 ). To offer a demonstration of an EAD application that can be utilized in conjunction with existing barcode-based and SNP-based multiplexing methods, we attempted to assign cells using Mixed-D for each singlet that was discordant across HTO demultiplexing tools. When studying a biological phenomenon, it is typical to use several complementary techniques (such as Western blotting and qPCR) to enhance the reliability of findings. If the results of different techniques are similar, there is a boost in confidence in the overall conclusion. By using complementary techniques, we can analyse cell assignments from various angles, such as barcode-based and expression-based, and the convergence of results indicates the reliability of the assignments. Mixed-D only approved singlets proposed by EAD ( T = 1) that were identical to the proposed donor by at least three tools in cellhashR output. The mixed-D was able to save 50.4% (993/1968) of discordant singlets (Min. 40.95%, Median 50.00%, Max. 59.05%) and it improved the HTO demultiplexing results by an average of 1.4% (993/70 000) (Min. 1.25%, Median 1.33%, Max. 1.74%). Since three HTO demultiplexing tools have previously verified the Mixed-D assignments and we employ a threshold equal to 1 for EAD, we can anticipate secure assignments for the discordant cells.
The cost of single-cell sequencing is influenced by various factors, such as the specific technology employed, the number of cells sequenced, the sequencing depth, and the provider. In most cases, the cost of scRNA-seq is ∼£1–5 per cell. Considering the dataset of blood and liver MAIT cells as an example, mixed demultiplexing can reduce the number of discordant cells (from 1968 to 975) and save around £3000. Expression-based demultiplexing is a promising complementary approach to identifying individual samples in a pooled sequencing experiment. By combining the results of both methods, we can obtain a more precise picture of the actual demultiplexing results.
Interindividual DCE analysis within the same cell type can be used to identify cells from the same donor in different activation states, as indicated by the results from the MAIT cell dataset ( Garner et al. 2023 ). The use of interindividual DCE analysis for scRNA-seq is not limited to EAD. It can potentially be applied to discover small changes in gene expression that are linked to various conditions or cell states. Researchers who want to study the transcriptional heterogeneity of cell states within the same cell type can benefit from this technique ( Garner et al. 2023 ). These applications are covered in more detail in the following section.
In this section, we show that it is possible to identify genes that are differentially co-expressed for each pair of donors for nonimmune cells. Using DCE pairs between individuals allows for EAD. Furthermore, we examine examples of variations in co-expressed gene patterns among various individuals that can shed light on the molecular mechanisms underlying differences in therapeutic response.
To demonstrate that our method can be utilized for nonimmune cells, we used single-nuclei transcriptome profiles of the SN and cortex (middle frontal gyrus) (in total 12 samples, including two SN replicates) ( Agarwal et al. 2020 ). The analysis of cell-type enrichment showed that there were five different cell populations in each sample of the SN and six different cell populations in the cortex (e.g. astrocytes) ( Agarwal et al. 2020 ). We were able to identify distinct clusters of cells across individuals based on the expression pattern of DCE genes despite cell heterogeneity ( Fig. 9 ). In at least one pair of samples, we found 3207 genes with DCE (FDR < 1 −10 ), with 568 of them being detected in only one pair of samples for the indicated region. We did not observe DCE patterns between replicated samples (FDR < 1 −4 ). Our achievement of finding representative DCE genes for each pair of donors allows us to use EAD for this data type.
A few examples of the many interindividual DCE genes that have been identified in the substantia nigra and cortex. These genes play important roles in a variety of neurological processes, and their dysregulation can contribute to the development of neurological disorders. More investigation is required to fully comprehend the roles of these genes in the brain and their potential as therapeutic targets for neurological disorders. (a, b) Examples of interindividual DCE pattern in the substantia nigra and cortex. (c) The t-SNE project of transcriptionally and functionally distinct clusters, highlighting microglia and neuron cell type groups, is presented. Pink/Red cells have passed the threshold of cell type enrichment ( Nassiri et al. 2023 ). (d) Examples of DCE patterns between cell types in the substantia nigra and cortex. (e) A pair of genes that exhibit differential co-expression but not differential expression. The analyses were carried out by employing single-cell expression profiles across the SN and cortex regions of a donor. In each region, the single-cell expression distributions of CTNNA3 and FAM221A genes are visualized by a violin plot. The expression levels of both genes decrease from SN to cortex, but only CTNNA3 has a significant differential expression. (f) The CTNNA3 and FAM221A gene pairs exhibit DCE patterns in the SN and cortex.
One example is the differential co-expression of MALAT1 and LINGO1 in the cortex of different individuals ( Fig. 9a ). MALAT1 is a long noncoding RNA (lncRNA) that is highly expressed in the brain, particularly in the cortex. MALAT1 has the potential to be a promising therapeutic target for a variety of neurological disorders ( Wang et al. 2022 ). Neuronal survival and axonal growth are regulated by LINGO1 , which is a leucine-rich single transmembrane protein, in the cortex. It has been observed that it promotes the death of neurons and hinders the growth of axons. It is suggested that LINGO1 may be involved in both the normal pruning of neurons during development and the pathological loss of neurons that occurs in neurodegenerative diseases ( Inoue et al. 2007 ). LINGO1 has the potential to be a promising therapeutic target for several neurological disorders. MALAT1 and LINGO1 have been demonstrated to interact and have opposite effects on neuronal survival ( Fan et al. 2018 ). Neuronal survival is promoted by MALAT1 , while neuronal death is promoted by LINGO1 ( Inoue et al. 2007 , Wang et al. 2022 ). The balance between MALAT1 and LINGO1 expression may play a significant role in regulating neuronal survival in the cortex. Further research is needed to fully understand the differential co-expression relationship between upregulation of MALAT1 and downregulation of LINGO1 , as well as its roles in neurological disorders.
Other examples of interindividual DCE pairs include XIST and ARL17B in SN ( Fig. 9b ). The SN expresses XIST , which has been shown to regulate the expression of genes crucial for dopaminergic neuron function ( Wang et al. 2021 ). ARL17B is present in the SN and has been discovered to play a significant role in the survival of dopaminergic neurons ( Reus et al. 2021 ). Parkinson’s disease, which is characterized by the loss of dopaminergic neurons in the SN, has been linked to mutations in XIST and ARL17B ( Wang et al. 2021 ). Our result indicates that there is a settled co-expression relationship between XIST and ARL17B . Further exploration is required to fully comprehend the functional consequence of losing association between XIST and ARL17B expression in SN and their possible contribution to Parkinson’s disease.
The expression patterns of many genes are not directly co-regulated within cell types, but they are differentially co-expressed across cell types, as shown by our results. For instance, the interaction between ARID2 and CNTNAP2 resolves in glial cells of the SN ( Fig. 9c and d ). This demonstrates differential co-expression between cortical neurons and SN glial cell types. The regulation of gene expression and DNA replication by chromatin remodelling is facilitated by ARID2 (AT-rich interaction domain 2) protein ( Kang et al. 2021 ). It is particularly important for glial cells to maintain homeostasis and protect against infection and neurodegeneration. In glial cells, CNTNAP2 (contactin-associated protein 2) is a key component of glial-neuronal communication, neuroprotection, and cognitive function ( Gandhi et al. 2023 , St George-Hyslop et al. 2023 ). The interaction between CNTNAP2 and ARID2 leads to the stabilization of the ARID2 protein ( Moffat et al. 2022 ). ARID2 needs this stabilization to properly regulate gene expression ( Moffat et al. 2022 ). Our results suggest that the co-expression of ARID2 and CNTNAP2 is resolved in the microglia and astrocyte cells in SN ( Fig. 9d ). Further research is needed to fully elucidate the effects of losing the interaction between ARID2 and CNTNAP2 on microglial and astrocyte cell functions in SN.
Our observation revealed a distinct set of genes that represent each brain region and cell-type. In addition, we found examples of interindividual variations in co-expressed gene patterns, which provide insight into the challenges presented by variation in therapeutic response ( Crowell et al. 2020 , Guo et al. 2023 ).
Typical differential gene expression analysis (DiffEx) does not provide a similar explicit model for co-expression relations between molecules at the single-cell level. The focus of DiffEx is on the expression levels of individual genes ( Gaublomme et al. 2019 ). Differential co-expression analysis (DiffCoEx) is focused on discovering genes whose expression levels change significantly under various conditions. DiffCoEx offers insight into the coordinated response of genes and assists us in understanding the underlying biological networks ( McKenzie et al. 2016 , Crowell et al. 2020 ). As a demonstration, we compare DiffEx and DiffCoEx results across SN and cortex regions in a 56-year-old female donor ( Fig. 9e and f ). Differential expression of genes across the SN and cortex was found by the Seurat tool, and 64.8% (5483/8456) of them were also differentially co-expressed (FDR<1 −3 ) ( Hao et al. 2024 ). The example of CTNNA3 and FAM221A genes illustrates the distinction between differential expression and differential co-expression. In both SN and cortex contexts, Fig. 9e illustrates where the CTNNA3 and FAM221A genes are expressed. The expression values of both genes decrease from SN to cortex, but FAM221A is not a significant differentially expressed gene ( P = 1), as shown in Fig. 9e . These genes have significant differential co-expression in the SN and cortex, which cannot be predicted solely based on differential expression relationships ( Fig. 9f ).
The capability of multiplex scRNA-seq samples has attracted research attention to lowering experiment costs and addressing batch effects. Several multiplexing methods and bioinformatics tools have been developed for demultiplexing pooled datasets ( Kang et al. 2018 , Stoeckius et al. 2018 , Gaublomme et al. 2019 , Guo et al. 2019 ). The main approaches include barcode-based ( Boggy et al. 2022 ) and SNP-based sample pool demultiplexing ( Kang et al. 2018 , Huang et al. 2019 ). Compared to other approaches, EAD facilitates feature selection (marker gene selection) for clustering single-cell data by obtaining individual-specific variability ( Ranjan et al. 2021 ). EAD has added benefits to previous methods by improving the accuracy of cell assignments into individual samples, without the need for additional experimental steps ( Nassiri et al. 2023 ).
The accurate assigning of cells to their respective donors can provide valuable insights for researchers into the biological processes and disease mechanisms that vary across individuals. The findings indicate that combining expression-based demultiplexing with SNP-based or barcode-based methods is the most accurate approach for demultiplexing single-cell and single-nuclei sequencing data. The combined demultiplexing results are more accurate due to the fact that they take into account the strengths of methods. For example, cell-multiplexing oligos using cells or nuclei samples can be technically challenging and have limited performance ( Stoeckius et al. 2018 ). On the other hand, the identity of donors (donor-specific information) is not specified by EAD. When cells are assigned to donor groups using EAD, the identity of donors can be specified by incorporating demultiplexing results provided by barcode-based multiplexing assays ( Kim et al. 2022 ).
Each person has a distinct set of SNPs. In the event that two cells from different individuals are combined to form a doublet, the combined SNP profile will appear unbalanced. SNP demultiplexing algorithms have the ability to identify these inconsistencies and label them as potential doublets ( Kang et al. 2018 , Huang et al. 2019 ). SNP demultiplexing for doublet detection is not perfect and can lead to false positives. Factors like sequencing depth and quality, and completeness of reference panels used for SNP calling determine the accuracy of doublet detection ( Kang et al. 2018 ). False positives can be caused by sequencing errors, allelic dropout, batch effects, somatic mutations, and natural variability in gene expression ( Huang et al. 2019 ). While expression-based demultiplexing cannot detect doublets, the combination of SNP and expression demultiplexing can offer more confirmation or reassign doublet predictions. Filtering out cells that are classified as true doublets using tools like SCAN is recommended before reassigning a predicted doublet using Mixed-D ( Dahlin et al. 2018 ).
The validation approach for characterization of co-expression variations is limited by the use of datasets from the 10× Genomics Chromium platform. In general, it is plausible that targeted scRNA-seq is less likely to obtain enough SNPs for SNP-based demultiplexing and has a limited number of genes for DCE analysis. By incorporating other single-cell RNA and protein expression technologies that have been developed, our approach can be easily enhanced.
The application of interindividual DCE analysis of scRNA-seq goes beyond EAD. It can potentially be applied to reveal biological activity that is useful for patient stratification, identifying biomarkers of target engagement, and connecting genomic programs with biological functions ( McKenzie et al. 2016 , Badia-i-Mompel et al. 2023 ). The study of interindividual variation in single-cell RNA-sequencing data is a constantly evolving field that involves continuous methodological advancement and refinement ( Kumasaka et al. 2023 ). Most single-cell studies employ average gene expression profiles across cell types or states of interest, which often obscures differences among individuals that are apparent at single-cell resolution ( Murdock and Tsai 2023 ). Variability between individuals remains an understudied aspect of relationships between molecules at the single-cell level ( Crowell et al. 2020 ). Cell-level mixed models are currently employed for differential state analysis across multiple samples (e.g. donors) and experimental conditions using cell-level measurements ( Crowell et al. 2020 ). The cell-level mixed models generate log-fold changes and the proportion of cells that have expressed a particular gene in each sample or group. These methods greatly underestimate the differences in gene expression between different cell subpopulations, such as those with low expression ( Crowell et al. 2020 , Auerbach et al. 2021 ). The cell-level solution to improve this issue is provided by DCE analysis, which directly models gene expression. In addition, it provides the possibility of modelling related gene regulatory networks and following each pair of DCE genes along the axis of samples. While the interindividual differential co-expression approach identifies how disease signatures vary between clinical subtypes, these changes ultimately might converge on shared signalling pathways that present biomarkers for new precision therapeutic strategies ( Johansen et al. 2023 ).
We anticipate that EAD will be applied to improve the results of SNP-based and barcode-based multiplexing assays, provide insight into challenges presented by variation in therapeutic response, and consider the effect of these stratification strategies on proposed candidate biomarker genes ( Stoeckius et al. 2018 , Li et al. 2020 ).
Isar Nassiri (Conceptualization [lead], Formal analysis [lead], Methodology [lead], Project administration [lead], Resources [lead], Software [lead], Validation [lead], Visualization [lead], Writing—original draft [lead], Writing—review & editing [lead]), Andrew J. Kwok (Data curation [supporting], Resources [supporting], Validation [supporting], Writing—original draft [supporting]), Aneesha Bhandari (Data curation [supporting], Resources [supporting], Validation [supporting], Writing—original draft [supporting]), Katherine R. Bull (Data curation [equal], Resources [equal], Validation [equal], Writing—original draft [equal]), Lucy C. Garner (Data curation [equal], Formal analysis [equal], Resources [equal], Validation [equal], Writing—original draft [equal], Writing—review & editing [equal]), Paul Klenerman (Data curation [equal], Resources [equal], Writing—review & editing [equal]), Caleb Webber (Data curation [equal], Resources [equal], Writing—review & editing [equal]), Laura Parkkinen (Conceptualization [equal], Data curation [equal], Resources [equal], Writing—review & editing [equal]), Angela W. Lee (Data curation [equal], Resources [equal], Writing—original draft [equal]), Yanxia Wu (Conceptualization [equal], Data curation [equal], Investigation [equal], Writing—original draft [supporting], Writing—review & editing [supporting]), Benjamin Fairfax (Data curation [equal], Methodology [supporting], Resources [equal], Writing—original draft [equal], Writing—review & editing [supporting]), Julian C. Knight (Investigation [equal], Resources [equal], Writing—original draft [equal], Writing—review & editing [equal]), David Buck (Conceptualization [equal], Resources [equal], Writing—original draft [equal]), and Paolo Piazza (Funding acquisition [lead], Methodology [equal], Project administration [supporting], Resources [equal], Validation [equal], Writing—original draft [equal], Writing—review & editing [equal])
Supplementary data are available at Bioinformatics Advances online.
None declared.
The research was supported by the Wellcome Trust Core Award [203141/Z/16/Z]. I.N. was supported by the National Institute for Health Research (NIHR) Oxford Health Biomedical Research Centre (BRC). The views expressed are those of the author(s) and not necessarily those of the NHS, the NIHR or the Department of Health.
Agarwal D , Sandor C , Volpato V et al. A single-cell atlas of the human substantia nigra reveals cell-specific pathways associated with neurological disorders . Nat Commun 2020 ; 11 : 4183 .
Google Scholar
Almeida A , Loy A , Hofmann H. ggplot2 compatible quantile-quantile plots in R . R J 2018 ; 10 : 248 – 61 .
Apostolidou S , Harbauer T , Lasch P et al. Fatal COVID-19 in a child with persistence of SARS-CoV-2 despite extensive multidisciplinary treatment: a case report . Children (Basel) 2021 ; 8 : 564 .
Auerbach BJ , Hu J , Reilly MP et al. Applications of single-cell genomics and computational strategies to study common disease and population-level variation . Genome Res 2021 ; 31 : 1728 – 41 .
Badia-I-Mompel P , Wessels L , Müller-Dott S et al. Gene regulatory network inference in the era of single-cell multi-omics . Nat Rev Genet 2023 ; 24 : 739 – 54 .
Beckelman BC , Day S , Zhou X et al. Dysregulation of elongation factor 1A expression is correlated with synaptic plasticity impairments in Alzheimer’s disease . J Alzheimers Dis 2016a ; 54 : 669 – 78 .
Beckelman BC , Zhou X , Keene CD et al. Impaired eukaryotic elongation factor 1A expression in Alzheimer’s disease . Neurodegener Dis 2016b ; 16 : 39 – 43 .
Boggy GJ , McElfresh GW , Mahyari E et al. BFF and cellhashR: analysis tools for accurate demultiplexing of cell hashing data . Bioinformatics 2022 ; 38 : 2791 – 801 .
Bongen E , Lucian H , Khatri A et al. Sex differences in the blood transcriptome identify robust changes in immune cell proportions with aging and influenza infection . Cell Rep 2019 ; 29 : 1961 – 73.e4 .
Cardiello JF , Joven Araus A , Giatrellis S et al. Evaluation of genetic demultiplexing of single-cell sequencing data from model species . Life Sci Alliance 2023 ; 6 : e202301979 .
Crowell HL , Soneson C , Germain P-L et al. Muscat detects subpopulation-specific state transitions from multi-sample multi-condition single-cell transcriptomics data . Nat Commun 2020 ; 11 : 6077 .
Dahlin JS , Hamey FK , Pijuan-Sala B et al. A single-cell hematopoietic landscape resolves 8 lineage trajectories and defects in kit mutant mice . Blood 2018 ; 131 : e1 – 11 .
Deelen P , Zhernakova DV , de Haan M et al. Calling genotypes from public RNA-sequencing data enables identification of genetic variants that affect gene-expression levels . Genome Med 2015 ; 7 : 30 .
Engebretsen S , Bohlin J. Statistical predictions with glmnet . Clin Epigenetics 2019 ; 11 : 123 .
Fairfax BP , Makino S , Radhakrishnan J et al. Genetics of gene expression in primary immune cells identifies cell type-specific master regulators and roles of HLA alleles . Nat Genet 2012 ; 44 : 502 – 10 .
Fairfax BP , Taylor CA , Watson RA et al. Peripheral CD8+ T cell characteristics associated with durable responses to immune checkpoint blockade in patients with metastatic melanoma . Nat Med 2020 ; 26 : 193 – 9 .
Fan B , Wei Z , Yao X et al. Microenvironment imbalance of spinal cord injury . Cell Transplant 2018 ; 27 : 853 – 66 .
Gandhi T , Canepa CR , Adeyelu TT et al. Neuroanatomical alterations in the CNTNAP2 mouse model of autism spectrum disorder . Brain Sci 2023 ; 13 : 891 .
Garner LC , Amini A , FitzPatrick MEB et al. Single-cell analysis of human MAIT cell transcriptional, functional and clonal diversity . Nat Immunol 2023 ; 24 : 1565 – 78 .
Gaublomme JT , Li B , McCabe C et al. Nuclei multiplexing with barcoded antibodies for single-nucleus genomics . Nat Commun 2019 ; 10 : 2907 .
Griffiths JA , Richard AC , Bach K et al. Detection and removal of barcode swapping in single-cell RNA-seq data . Nat Commun 2018 ; 9 : 2667 .
Grun B , Leisch F. FlexMix version 2: finite mixtures with concomitant variables and varying and constant parameters . J Stat Soft 2008 ; 28 : 1 – 35 .
Guo C , Kong W , Kamimoto K et al. CellTag indexing: genetic barcode-based sample multiplexing for single-cell genomics . Genome Biol 2019 ; 20 : 90 .
Guo MG , Reynolds DL , Ang CE et al. Integrative analyses highlight functional regulatory variants associated with neuropsychiatric diseases . Nat Genet 2023 ; 55 : 1876 – 91 .
Hao Y , Stuart T , Kowalski MH et al. Dictionary learning for integrative, multimodal and scalable single-cell analysis . Nat Biotechnol 2024 ; 42 : 293 – 304 .
Heaton H , Talman AM , Knights A et al. Souporcell: robust clustering of single-cell RNA-seq data by genotype without reference genotypes . Nat Methods 2020 ; 17 : 615 – 20 .
Howitt G , Feng Y , Tobar L et al. Benchmarking single-cell hashtag oligo demultiplexing methods . NAR Genom Bioinform 2023 ; 5 : lqad086 .
Huang M , Wang J , Torre E et al. SAVER: gene expression recovery for single-cell RNA sequencing . Nat Methods 2018 ; 15 : 539 – 42 .
Huang X , Huang Y. Cellsnp-lite: an efficient tool for genotyping single cells . Bioinformatics 2021 ; 37 : 4569 – 71 .
Huang Y , McCarthy DJ , Stegle O. Vireo: Bayesian demultiplexing of pooled single-cell RNA-seq data without genotype reference . Genome Biol 2019 ; 20 : 273 .
Inoue H , Lin L , Lee X et al. Inhibition of the leucine-rich repeat protein LINGO-1 enhances survival, structure, and function of dopaminergic neurons in Parkinson’s disease models . Proc Natl Acad Sci USA 2007 ; 104 : 14430 – 5 .
Johansen N , Somasundaram S , Travaglini KJ et al. Interindividual variation in human cortical cell type abundance and expression . Science 2023 ; 382 : eadf2359 .
Kang E , Kang M , Ju Y et al. Association between ARID2 and RAS-MAPK pathway in intellectual disability and short stature . J Med Genet 2021 ; 58 : 767 – 77 .
Kang HM , Subramaniam M , Targ S et al. Multiplexed droplet single-cell RNA-sequencing using natural genetic variation . Nat Biotechnol 2018 ; 36 : 89 – 94 .
Kim H-J , Booth G , Saunders L et al. Nuclear oligo hashing improves differential analysis of single-cell RNA-seq . Nat Commun 2022 ; 13 : 2666 .
Kumasaka N , Rostom R , Huang N et al. Mapping interindividual dynamics of innate immune response at single-cell resolution . Nat Genet 2023 ; 55 : 1066 – 75 .
Kwok AJ , Allcock A , Ferreira RC et al. ; Emergency Medicine Research Oxford (EMROx) . Neutrophils and emergency granulopoiesis drive immune suppression and an extreme response endotype during sepsis . Nat Immunol 2023 ; 24 : 767 – 79 .
Li B , Gould J , Yang Y et al. Cumulus provides cloud-based data analysis for large-scale single-cell and single-nucleus RNA-seq . Nat Methods 2020 ; 17 : 793 – 8 .
Li H. A statistical framework for SNP calling, mutation discovery, association mapping and population genetical parameter estimation from sequencing data . Bioinformatics 2011 ; 27 : 2987 – 93 .
Li H , Durbin R. Fast and accurate short read alignment with Burrows-Wheeler transform . Bioinformatics 2009 ; 25 : 1754 – 60 .
Lotta LA , Pietzner M , Stewart ID et al. ; MacTel Consortium . A cross-platform approach identifies genetic regulators of human metabolism and health . Nat Genet 2021 ; 53 : 54 – 64 .
Manach C , Milenkovic D , Van de Wiele T et al. Addressing the inter-individual variation in response to consumption of plant food bioactives: towards a better understanding of their role in healthy aging and cardiometabolic risk reduction . Mol Nutr Food Res 2017 ; 61 : 1600557 .
McCarthy DJ , Campbell KR , Lun ATL et al. Scater: pre-processing, quality control, normalization and visualization of single-cell RNA-seq data in R . Bioinformatics 2017 ; 33 : 1179 – 86 .
McGinnis CS , Patterson DM , Winkler J et al. MULTI-seq: sample multiplexing for single-cell RNA sequencing using lipid-tagged indices . Nat Methods 2019 ; 16 : 619 – 26 .
McKenzie AT , Katsyv I , Song W-M et al. DGCA: a comprehensive R package for differential gene correlation analysis . BMC Syst Biol 2016 ; 10 : 106 .
Moffat JJ , Smith AL , Jung E-M et al. Neurobiology of ARID1B haploinsufficiency related to neurodevelopmental and psychiatric disorders . Mol Psychiatry 2022 ; 27 : 476 – 89 .
Murdock MH , Tsai LH. Insights into Alzheimer’s disease from single-cell genomic approaches . Nat Neurosci 2023 ; 26 : 181 – 95 .
Nassiri I , Fairfax B , Lee A et al. scQCEA: a framework for annotation and quality control report of single-cell RNA-sequencing data . BMC Genomics 2023 ; 24 : 381 .
Nassiri I , McCall MN. Systematic exploration of cell morphological phenotypes associated with a transcriptomic query . Nucleic Acids Res 2018 ; 46 : e116 .
Neavin D , Senabouth A , Arora H et al. Demuxafy: improvement in droplet assignment by integrating multiple single-cell demultiplexing and doublet detection methods. Genome Biol 2024 ; 25 :94.
Oelen R , de Vries DH , Brugge H et al. ; BIOS Consortium . Single-cell RNA-sequencing of peripheral blood mononuclear cells reveals widespread, context-specific gene expression regulation upon pathogenic exposure . Nat Commun 2022 ; 13 : 3267 .
Ranjan B , Sun W , Park J et al. DUBStepR is a scalable correlation-based feature selection method for accurately clustering single-cell data . Nat Commun 2021 ; 12 : 5849 .
Reus LM , Pasaniuc B , Posthuma D et al. ; International FTD-Genomics Consortium . Gene expression imputation across multiple tissue types provides insight into the genetic architecture of frontotemporal dementia and its clinical subtypes . Biol Psychiatry 2021 ; 89 : 825 – 35 .
Rizzo AN , Yuan JXJ. NEDD9 provides mechanistic insight into the coagulopathy of COVID-19 . Pulm Circ 2022 ; 12 : e12087 .
St George-Hyslop F , Haneklaus M , Kivisild T et al. Loss of CNTNAP2 alters human cortical excitatory neuron differentiation and neural network development . Biol Psychiatry 2023 ; 94 : 780 – 91 .
Stoeckius M , Zheng S , Houck-Loomis B et al. Cell hashing with barcoded antibodies enables multiplexing and doublet detection for single cell genomics . Genome Biol 2018 ; 19 : 224 .
Tang L , Zhang D , Han P et al. Structural basis of SARS-CoV-2 and its variants binding to intermediate horseshoe bat ACE2 . Int J Biol Sci 2022 ; 18 : 4658 – 68 .
Van der Auwera GA , Carneiro MO , Hartl C et al. From FastQ data to high-confidence variant calls: the genome analysis toolkit best practices pipeline . Curr Protoc Bioinformatics 2013 ; 43 : 11.10.11 – .10.33 .
van der Wijst MGP , Brugge H , de Vries DH et al. ; BIOS Consortium . Single-cell RNA sequencing identifies celltype-specific cis-eQTLs and co-expression QTLs . Nat Genet 2018 ; 50 : 493 – 7 .
Wang L , Li S , Stone SS et al. The role of the lncRNA MALAT1 in neuroprotection against hypoxic/ischemic injury . Biomolecules 2022 ; 12 : 146 .
Wang W , Min L , Qiu X et al. Biological function of long non-coding RNA (LncRNA) xist . Front Cell Dev Biol 2021 ; 9 : 645647 .
Wolock SL , Lopez R , Klein AM. Scrublet: computational identification of cell doublets in single-cell transcriptomic data . Cell Syst 2019 ; 8 : 281 – 91.e9 .
Wu TZ et al. clusterProfiler 4.0: a universal enrichment tool for interpreting omics data . Innovation (Camb) 2021 ; 2 : 100141 .
Xin H , Lian Q , Jiang Y et al. GMM-Demux: sample demultiplexing, multiplet detection, experiment planning, and novel cell-type verification in single cell sequencing . Genome Biol 2020 ; 21 : 188 .
Xu J , Falconer C , Nguyen Q et al. Genotype-free demultiplexing of pooled single-cell RNA-seq . Genome Biol 2019 ; 20 : 290 .
Yang PY , Huang H , Liu CL. Feature selection revisited in the single-cell era . Genome Biol 2021 ; 22 : 321 .
Yazar S , Alquicira-Hernandez J , Wing K et al. Single-cell eQTL mapping identifies cell type-specific genetic control of autoimmune disease . Science 2022 ; 376 : eabf3041 .
Month: | Total Views: |
---|---|
June 2024 | 234 |
Citing articles via.
Oxford University Press is a department of the University of Oxford. It furthers the University's objective of excellence in research, scholarship, and education by publishing worldwide
Sign In or Create an Account
This PDF is available to Subscribers Only
For full access to this pdf, sign in to an existing account, or purchase an annual subscription.
New citation alert added.
This alert has been successfully added and will be sent to:
You will be notified whenever a record that you have chosen has been cited.
To manage your alert preferences, click on the button below.
Please log in to your account
Bibliometrics & citations, view options, recommendations, multi-view clustering via spectral partitioning and local refinement.
A new multi-view clustering algorithm is proposed.The proposed MVNC algorithm uses spectral partitioning and local refinement.MVNC is compared to state-of-the-art algorithms using three real-world datasets.MVNC significantly outperforms the other ...
Multi-view clustering has become a widely studied problem in the area of unsupervised learning. It aims to integrate multiple views by taking advantages of the consensus and complimentary information from multiple views. Most of the existing works in ...
Multi-view clustering methods based on tensor have achieved favorable performance thanks to the powerful capacity of capturing the high-order correlation hidden in multi-view data. However, many existing works only pay attention to exploring the ...
Published in.
Elsevier Science Inc.
United States
Author tags.
Other metrics, bibliometrics, article metrics.
Login options.
Check if you have access through your login credentials or your institution to get full access on this article.
Share this publication link.
Copying failed.
Affiliations, export citations.
We are preparing your search results for download ...
We will inform you here when the file is ready.
Your file of search results citations is now ready.
Your search export query has expired. Please try again.
IMAGES
VIDEO
COMMENTS
Browse various samples of academic assignments written by professional writers for different disciplines and levels. Learn how to plan, structure, and write your own assignment with our tips and guidance.
Find sample papers formatted in seventh edition APA Style for different types of professional and student papers. Download Word files to use as templates and edit them as needed for your own papers.
Assignment is a task given to students by a teacher or professor, usually as a means of assessing their understanding and application of course material. Assignments can take various forms, including essays, research papers, presentations, problem sets, lab reports, and more. Assignments are typically designed to be completed outside of class ...
Learn how to read and interpret college writing assignments by identifying the task, audience, evidence, style, and format. Find tips and examples for different types of assignments and key terms.
This handout provides information about annotated bibliographies in MLA, APA, and CMS. These OWL resources will help you understand and complete specific types of writing assignments, such as annotated bibliographies, book reports, and research papers. This section also includes resources on writing academic proposals for conference ...
Media Files: APA Sample Student Paper , APA Sample Professional Paper This resource is enhanced by Acrobat PDF files. Download the free Acrobat Reader. Note: The APA Publication Manual, 7 th Edition specifies different formatting conventions for student and professional papers (i.e., papers written for credit in a course and papers intended for scholarly publication).
In this sample paper, we've put four blank lines above the title. Commented [AF3]: Authors' names are written below the title, with one double-spaced blank line between them. Names should be written as follows: First name, middle initial(s), last name. Commented [AF4]: Authors' affiliations follow immediately after their names.
Come up with a thesis. Create an essay outline. Write the introduction. Write the main body, organized into paragraphs. Write the conclusion. Evaluate the overall organization. Revise the content of each paragraph. Proofread your essay or use a Grammar Checker for language errors. Use a plagiarism checker.
To construct an assignment structure, use outlines. These are pieces of text that relate to your topic. It can be ideas, quotes, all your thoughts, or disparate arguments. Type in everything that you think about. Separate thoughts scattered across the sheets of Word will help in the next step. Then it is time to form the text.
Popular essay samples. Though we have an extensive range of samples and are always adding and writing more, you'll find . a selection of the topics that students search for most often down below. Search. Social Issues: Animal Rights. Teenage Pregnancy. Сyber Bullying. Racism.
View or download authentic sample assignments for academic writing skills. Find annotated bibliography, essay, literature review, reflective writing and report examples.
Learn how to write effective assignments with clear and simple tips on outline, introduction, structure, referencing, conclusion and more. See an example of an assignment on obesity epidemic and its causes and solutions.
Tables and figures are only included in student papers if needed for the assignment. Tables and figures share the same elements and layout. See the website for sample tables and sample figures. Table elements. Tables include the following four elements: Number; Title; Body (rows and columns) Note (optional if needed to explain elements in the ...
hardest thinking, and feel the greatest sense of mastery and. growth, in their writing. Cour. es. and assignments should be planned with this in mi. d. Three principles are paramount:1. Name what you want and imagine students doing itHowever free students are to range and explore in a paper, the general kind of paper you're inviting has com.
nment. Some people find it easiest to write the introduction first, whereas others leave it until. the end. Neither approach is right or wrong, so write the assignment in whichever order feels best. for you. The introduction might be up to around 10% of the word count (e.g. up to 200 words for a 2000 word ass.
A sample of a business-style report with an annotated format. Investigative report sample. Sample of an investigative report [PDF 500KB]. A resource for science, engineering and technology students. How to write an investigative report, including an annotated format. Assignment topics and editing. Interpreting assignment topics [PDF 370 KB ...
This guide includes tips on writing common course assignments. Both in traditional and online classrooms, journal entries are used as tools for student reflection. By consciously thinking about and comparing issues, life experiences, and course readings, students are better able to understand links between theory and practice and to generate ...
From college assignment samples to university essay assignment samples, students can access all kinds of resources from our collection. Our top-qualified experts curate all these solved papers. So, if you check a sample of our essay writing, you can immediately notice the difference in quality between us and a fake. We also regularly update our ...
Learn how to create effective writing assignments for your course by defining the purpose, audience, mode, and criteria. Find tips on sequencing, checking, and selecting the format of your assignments.
43 Sample Writing Assignments Process Analysis. Be the expert and teach your reader how to do something. You will focus on writing the main steps to completing this process and organizing based on chronology or priority of steps. This essay could be demonstrative in nature (ex. How to bathe and groom a dog at home or how to make banana nut ...
Low stakes writing helps: Describe, apply, and retain information. Explore and personalize ideas. Focus thoughts and questions. Demonstrate the value of writing as a part of the learning process. Informally engage each student in the classroom. Improve high-stakes writing. Efficiently assess student learning.
This guide will provide research and writing tips to help students complete a literature review assignment. Home; Steps for Conducting a Lit Review; ... Chicago: Notes Bibliography This link opens in a new window; MLA Style This link opens in a new window; Sample Literature Reviews. Sample Lit Reviews from Communication Arts; Have an exemplary ...
Assignment and Course Assessment KCTL Spring Workshop 5/2/2024. Today's Objectives •Explain what rubrics are and how they differ from ... Using samples of existing work, divide into levels corresponding to the levels you set. What are the qualities of the best work? The
Lecture/text homework assignment # 6 Note: Please circle your answers when appropriate! 0) Problems 6 & 7 from last time are due; note that your recitation instructor may ask you to present ... For both the sample standard deviation is about s = 7.65 cm. (a) Suppose you take a sample of 8 women and 6 men. Construct a 99% CI for both. Do the
Cite your source automatically in MLA or APA format. Cite. Using citation machines responsibly. Powered by. Media File: APA Sample Paper. This resource is enhanced by an Acrobat PDF file. Download the free Acrobat Reader. Click this link to download the PDF handout of the APA Sample Paper.
An assignment can be submitted late, with a participation-score penalty of 3 points per 24-hour-period, or fraction thereof, after the deadline. The teaching staff is happy to discuss assignment requirements, and to discuss how you might go about completing them, but cannot confirm in advance that a draft assignment fulfills the requirements.
We utilized 30 samples from both sepsis and healthy individuals in six batches to assess the performance of classification approaches. The results indicate that combining genetic and EAD results can enhance the accuracy of assignments (Min. 0.94, Mean 0.98, Max. 1).
Finally, a sample assignment matrix is learned via jointly reconstructing the samples using the two bases further cooperating with spectral clustering. In this way, the heterogeneity gap is bridged by tensor feature matching, and the inaccurate sample-wise similarity estimation is omitted by using feature-wise matching to guide sample-wise ...
login.microsoftonline.com