Abstract

We present pipeComp ( https://github.com/plger/pipeComp ), a flexible R framework for pipeline comparison handling interactions between analysis steps and relying on multi-level evaluation metrics. We apply it to the benchmark of single-cell RNA-sequencing analysis pipelines using simulated and real datasets with known cell identities, covering common methods of filtering, doublet detection, normalization, feature selection, denoising, dimensionality reduction, and clustering. pipeComp can easily integrate any other step, tool, or evaluation metric, allowing extensible benchmarks and easy applications to other fields, as we demonstrate through a study of the impact of removal of unwanted variation on differential expression analysis. Document type: Article

Abstract

Background Chromatin accessibility profiling assays such as ATAC-seq and DNase1-seq offer the opportunity to rapidly characterize the regulatory state of the genome at a single nucleotide resolution. Optimization of molecular protocols has enabled the molecular biologist to produce next-generation sequencing libraries in several hours, leaving the analysis of sequencing data as the primary obstacle to wide-scale deployment of accessibility profiling assays. To address this obstacle we have developed an optimized and efficient pipeline for the analysis of ATAC-seq and DNase1-seq data. Results We executed a multi-dimensional grid-search on the NIH Biowulf supercomputing cluster to assess the impact of parameter selection on biological reproducibility and ChIP-seq recovery by analyzing 4560 pipeline configurations. Our analysis improved ChIP-seq recovery by 15% for ATAC-seq and 3% for DNase1-seq and determined that PCR duplicate removal improves biological reproducibility by 36% without significant costs in footprinting transcription factors. Our analyses of down sampled reads identified a point of diminishing returns for increased library sequencing depth, with 95% of the ChIP-seq data of a 200 million read footprinting library recovered by 160 million reads. Conclusions We present optimized ATAC-seq and DNase-seq pipelines in both Snakemake and bash formats as well as optimal sequencing depths for ATAC-seq and DNase-seq projects. The optimized ATAC-seq and DNase1-seq analysis pipelines, parameters, and ground-truth ChIP-seq datasets have been made available for deployment and future algorithmic profiling. Electronic supplementary material The online version of this article (10.1186/s12864-018-4943-z) contains supplementary material, which is available to authorized users. Document type: Article