GitHub
10/22/2026, 7:54 AMauto, for a pipeline-level test of strandedness inference in nf-core/rnasplice (nf-core/rnasplice#338, following nf-core/rnasplice#336). The test is added in nf-core/rnasplice#339.
All the existing GEUVADIS test reads are unstranded. The new files are derived from them: where read 1 of a pair aligns to the same strand as the transcripts it overlaps, the two mates are swapped, so read 1 ends up antisense as in a dUTP library. Pairs that overlap transcripts on both strands, or no transcript, are unchanged. Each sample keeps its 50,000 pairs and is about the size of the original files (8 files, about 29 MB in total).
• `testdata/*_chrX_reverse_{1,2}.fastq.gz`: the derived reads
• `samplesheet/samplesheet_strandedness_auto.csv`: ERR188383 and ERR188454 set to auto, the other two to reverse
• `scripts/make_reverse_stranded.py`: the script used, with the commands in the README
Checked with nf-core/rnasplice dev (-profile test,docker --strand_check and this samplesheet): both auto samples are called reverse (Salmon about 96% antisense, RSeQC about 91%), the strand check passes for all four samples, and featureCounts, DEXSeq, rMATS and Salmon run with reverse stranded settings.
Generated by Claude Opus 5.5
🤖 Generated with Claude Code
https://claude.ai/code/session_019Ynk7qHMBazfbYEBMLz4wd
nf-core/test-datasetsGitHub
10/22/2026, 8:29 AMGitHub
10/22/2026, 8:30 AM