Biology subjects

Freytag, S.

Publications and source records attributed to Freytag, S..

3 recordsLinked to original sources

bioRxiv2018

scRNA-seq mixology: towards better benchmarking of single cell RNA-seq protocols and analysis methods

Single cell RNA sequencing (scRNA-seq) technology has undergone rapid development in recent years, bringing with it new challenges in data processing and analysis. This has led to an explosion of tailored analysis methods for scRNA-seq to address various biological questions. However, the current lack of gold-standard benchmarking datasets makes it difficult for researchers to evaluate the performance of the many methods. Here, we designed and carried out a realistic benchmark experiment that included mixtures of single cells or pseudo-cells created by sampling admixtures of cells or RNA from 3 distinct cancer cell lines. Altogether we generated 10 datasets using a combination of droplet and plate-based scRNA-seq protocols, with varying data quality, population heterogeneity and noise levels. Using these benchmark datasets, we compared different protocols, evaluated the spike-in standard and multiple data analysis methods for tasks ranging from normalization and imputation, to clustering, trajectory analysis and data integration. Evaluation of methods across multiple datasets revealed some that performed well in general and others that suited specific situations. Our dataset and analysis provide a comprehensive comparison framework for benchmarking most popular scRNA-seq analysis tasks.

bioinformatics↗

bioRxiv2018

dtangle: accurate and fast cell-type deconvolution

MotivationUnderstanding cell type composition is important to understanding many biological processes. Furthermore, in gene expression studies cell type composition can confound differential expression analysis (DEA). To aid understanding cell type composition, methods of estimating (deconvolving) cell type proportions from gene expression data have been developed.\n\nResultsWe propose dtangle, a new cell-type deconvolution method. dtangle works on a range of DNA microarray and bulk RNA-seq platforms. It estimates cell-type proportions using publicly available, often cross-platform, reference data. To comprehensively evaluate dtangle, we assemble ten benchmark data sets. Here, dtangle is competitive with published deconvolution methods, is robust to selection of tuning parameters and is quicker than other methods. As a case study, we investigate the human immune response to Lyme disease. dtangles estimates reveal a temporal trend consistent with previous findings and are important covariates for DEA across disease status.\n\nAvailabilitydtangle is on CRAN (cran.r-project.org/package=dtangle) or github (dtangle.github.io).\n\nContactgjhunt@umich.edu

bioinformatics↗

bioRxiv2017

Cluster Headache: Comparing Clustering Tools for 10X Single Cell Sequencing Data

The commercially available 10X Genomics protocol to generate droplet-based single cell RNA-seq (scRNA-seq) data is enjoying growing popularity among researchers. Fundamental to the analysis of such scRNA-seq data is the ability to cluster similar or same cells into non-overlapping groups. Many competing methods have been proposed for this task, but there is currently little guidance with regards to which method offers most accuracy. Answering this question is complicated by the fact that 10X Genomics data lack cell labels that would allow a direct performance evaluation. Thus in this review, we focused on comparing clustering solutions of a dozen methods for three datasets on human peripheral mononuclear cells generated with the 10X Genomics technology. While clustering solutions appeared robust, we found that solutions produced by different methods have little in common with each other. They also failed to replicate cell type assignment generated with supervised labeling approaches. Furthermore, we demonstrate that all clustering methods tested clustered cells to a large degree according to the amount of genes coding for ribosomal protein genes in each cell.

bioinformatics↗