Fig. 9
From: recount3: summaries and queries for large-scale RNA-seq expression and splicing
Monorail aggregation workflow. The workflow runs on a single computer, taking all run-level analysis outputs as its input. Its outputs are the study- and collection-level objects and indexes hosted by recount3 and Snaptron. The disjoint exonic interval summaries furnish raw counts that can be summed into both exon- and gene-level counts. The “add motifs” step for junctions adds the specific donor and acceptor dinucleotide patterns present in the reference in genome for that junction. For the “add annotation status” step for junctions, an inclusive set of gene annotations is used, as detailed in Reference files. Not pictured: metadata is also aggregated and hosted by recount3 and Snaptron. Also not pictured: bigWig files are organized into per-study subdirectories on the host filesystem. See also Fig. 1 and Monorail details