SPLASH2 provides ultra-efficient, scalable, and unsupervised discovery on raw sequencing reads

Abstract

NOMAD is a new, unsupervised, reference-free, and unifying algorithm that discovers regulated sequence variation through statistical analysis of k-mer composition in DNA or RNA sequencing experiments. It subsumes many application-specific algorithms, from splicing detection to RNA editing to applications in DNA-sequencing and beyond. Here, we introduce NOMAD2, a fast, scalable, and user-friendly implementation of NOMAD based on KMC, an efficient k-mer counting approach. The pipeline has minimal installation requirements, and can be executed with a single command. NOMAD2 enables efficient analysis of massive RNA-Seq datasets where it reveals novel biology, showcased by rapid analysis of 1,553 human muscle cells, the entire Cancer Cell Line Encyclopedia (671 cell lines, 5.7 TB) and a deep RNAseq study of Amyotrophic Lateral Sclerosis (ALS) with ~2 fold less computational resource and time than state of the art alignment methods. NOMAD2 enables reference-free biological discovery at unmatched scale and speed. By bypassing genome alignment, we provide examples of its new insights into RNA expression in normal and disease tissue, to introduce NOMAD2 to enable expansive biological discovery not previously possible.

A reference-free algorithm discovers regulation in the plant transcriptome

“…The implementation of SPLASH used in this paper is described in 8 . Specifically, SPLASH version 1.9.0 (https://github.com/refresh-bio/SPLASH/tree/archive/1.9.0) was run on each dataset.…”

“…These problems, while acute in plants, are a general obstacle to discovery across the tree of life. To address them, we recently introduced a new approach to analyze regulation of genomes and transcriptomes using an ultra-efficient, reference-free, statistical approach called SPLASH 7,8 . In brief, SPLASH identifies statistically significant sample-specific sequence variation directly from raw reads, bypassing alignment to a reference 7,8 .…”

“…To address them, we recently introduced a new approach to analyze regulation of genomes and transcriptomes using an ultra-efficient, reference-free, statistical approach called SPLASH 7,8 . In brief, SPLASH identifies statistically significant sample-specific sequence variation directly from raw reads, bypassing alignment to a reference 7,8 . SPLASH processes raw FASTQ reads and records sequences of a set length (i.e.…”