Sign up to save your podcasts
Or
Share Conditional resampling improves sensitivity and specificity of single cell CRISPR regulatory screens
Share to email
Share to Facebook
Share to X
Share to email
Share to Facebook
Share to X
Or
Conditional resampling improves sensitivity and specificity of single cell CRISPR regulatory screens
Link to bioRxiv paper:
http://biorxiv.org/cgi/content/short/2020.08.13.250092v1?rss=1
Authors: Katsevich, E., Roeder, K.
Abstract:
Mapping gene-enhancer regulatory relationships is key to unraveling molecular disease mechanisms based on GWAS associations in non-coding regions. Recently developed CRISPR regulatory screens (CRSs) based on single cell RNA-seq (scRNA-seq) are a promising high-throughput experimental approach to this problem. However, the analysis of these screens presents significant statistical challenges, including modeling cell-level gene expression and correcting for sequencing depth. Using a recent large-scale CRS and its original analysis as a case study, we demonstrate weaknesses in existing analysis methodology, which lead to false positives as well as false negatives. To address these challenges, we propose SCEPTRE: analysis of single cell perturbation screens via conditional resampling. This novel method infers gene-enhancer associations by modeling the stochastic assortment of CRISPR gRNAs among cells instead of the gene expression, remaining valid despite arbitrary misspecification of the gene expression model. Applying SCEPTRE to the large-scale CRS, we demonstrate improvements in both sensitivity and specificity. We also discover 217 regulatory relationships not found in the original study, many of which are supported by existing functional data.
Copy rights belong to original authors. Visit the link for more info
View all episodes
Link to bioRxiv paper:
http://biorxiv.org/cgi/content/short/2020.08.13.250092v1?rss=1
Authors: Katsevich, E., Roeder, K.
Abstract:
Mapping gene-enhancer regulatory relationships is key to unraveling molecular disease mechanisms based on GWAS associations in non-coding regions. Recently developed CRISPR regulatory screens (CRSs) based on single cell RNA-seq (scRNA-seq) are a promising high-throughput experimental approach to this problem. However, the analysis of these screens presents significant statistical challenges, including modeling cell-level gene expression and correcting for sequencing depth. Using a recent large-scale CRS and its original analysis as a case study, we demonstrate weaknesses in existing analysis methodology, which lead to false positives as well as false negatives. To address these challenges, we propose SCEPTRE: analysis of single cell perturbation screens via conditional resampling. This novel method infers gene-enhancer associations by modeling the stochastic assortment of CRISPR gRNAs among cells instead of the gene expression, remaining valid despite arbitrary misspecification of the gene expression model. Applying SCEPTRE to the large-scale CRS, we demonstrate improvements in both sensitivity and specificity. We also discover 217 regulatory relationships not found in the original study, many of which are supported by existing functional data.
Copy rights belong to original authors. Visit the link for more info