LineageProfiler: Automated Classification and Visualization of Cell Type Identity for Mammalian Transcriptomes

摘要

Both microarray and next generation RNA sequencing methods have vastly improved our ability to detect transcript variation underlying organism development and disease. While many tools exist to assess gene and transcript variation, there is a paucity of methods to evaluate cell type identity relative to the hundreds of known adult and progenitor cell types. Such methods are sorely needed to understand which cell types are present within a biological sample, particularly during lineage restricted in vitro stem cell differentiation. We have developed LineageProfiler as a component of the AltAnalyze analysis package (http://www.altanalyze.org), to analyze and visualize transcriptome correlations to a large compendium of tissues, isolated cell types or progenitor states. Unlike related methods, LineageProfiler can utilize gene or exon expression profiles from either microarray or next generation sequencing data to derive correlations. Associated Z scores are automatically visualized along a comprehensive lineage network or as a clustered heatmap. Through integration with the tool GO-Elite (http://www.genmapp.org/go_elite), underlying biomarkers are used to evaluate enrichment of cell types between conditions and samples. This approach has been successful at accurately identifying known populations of differentiating cells in vitro from RNA-Seq, measuring the relative abundance of cell types from mixed tissue experiments and identifying contamination due to inconsistent tissue dissection.