<#2346 rnasplice: add a chrX subset of the minimal...
# test-data
g
#2346 rnasplice: add a chrX subset of the minimal test dataset Pull request opened by piplus2 Adds
chrX_subset/
, a smaller version of the nf-core/rnasplice minimal test dataset, to cut the disk, memory and time its CI needs. Every file is under 1 MB (20 MB in all). The minimal tests build their STAR and Salmon indices from all of chrX (156 Mb, 2,392 genes), although most reads fall on a few hundred genes. On the nf-core/rnasplice CI this means a 1.7 GB STAR index and 0.9 GB of Salmon index per test, and STAR needs 4.2 GB to build the index and 4.9 GB to align (see nf-core/rnasplice#332, which raised the runner volume to 80 GB). What is in
chrX_subset/
• `reference/X.fa.gz`: the regions of chrX around the 30 genes with at least two transcripts and the most reads across the four samples, plus the three genes the pipeline plots with MISO by default. Each gene is padded by 1 kb, regions closer than 10 kb are merged and widened until no gene crosses their ends, and the 30 regions are joined into one 3.4 Mb contig, still named
X
, separated by 1 kb of
N
. • `reference/genes_chrX.gtf`: the 82 genes inside the regions, in the new coordinates, with only the
transcript
,
exon
and
CDS
lines and the attributes the pipeline uses.
genes_chrX.gff3
and
misopy_index.tar.gz
are made from it as before. • `testdata/*_chrX_{1,2}.fastq.gz`: 10,000 pairs per sample, drawn with a fixed seed from the pairs whose two mates fall in the same region. • BAMs (4,000 of those pairs per sample, all with a spliced mate, to stay under 1 MB and still give the splicing tools junctions), Salmon results, DRIMSeq and DEXSeq inputs and the reverse stranded reads, regenerated from a pipeline run on this reference, with the same names and layout as the current files. • `samplesheet/`: the minimal samplesheets, pointing at
chrX_subset/
. •
scripts/make_chrx_subset.py
,
scripts/filter_pairs.py
and
scripts/package_chrx_subset.sh
, with the commands in the README. The current files are unchanged, so released versions of the pipeline keep working. Effect Running
-profile test
with every analysis (plus LeafCutter) on this data: | chrX | chrX_subset | | | ---------------------- | ----------- | ---------- | | STAR index | 1.7 GB | 42 MB | | Salmon index | 0.9 GB | 23 MB | | Work directory | 3.4 GB | 0.17 GB | | STAR align, per sample | 4.9 GB | under 1 GB | | Sum of task run times | 12.4 min | 5.4 min | Every analysis still reports results (for example 21 rMATS skipped exon events, 44 DEXSeq genes, 338 SUPPA events, 3 LeafCutter clusters), and so do the
genome_bam
and
transcriptome_bam
tests on the smaller BAMs. The pipeline PR, which points the tests at this folder and updates their snapshots, will follow once this is merged. Generated by Claude Opus 5.5 🤖 Generated with Claude Code https://claude.ai/code/session_019Ynk7qHMBazfbYEBMLz4wd nf-core/test-datasets
Review on #2346 rnasplice: add a chrX subset of the minimal test dataset Review approved by mashehu on piplus2's pull request nf-core/test-datasets