Add gedi/price test data: chr19+chr22 Ribo-seq cohort

README.md

@@ -458,6 +458,11 @@ The earth sciences folder contain subfolders for different data formats encounte
       - plastid
         - Homo_sapiens.GRCh38.111_chr20_rois.txt: metagene generated from Homo_sapiens.GRCh38.111_chr20.gtf using plastid `metagene generate` command
         - SRX11780887_p_offsets.txt: p-site offsets genereated from SRX11780887_chr20.bam and Homo_sapiens.GRCh38.111_chr20.gtf using plastid `psite` command
+      - price
+        - Homo_sapiens.GRCh38_chr19_22.pc_exon_masked.fa.gz: chr19+chr22 of Ensembl GRCh38 primary assembly, hard-masked outside protein-coding exons, for testing gedi/price (PRICE requires ~2x the candidate-ORF count chr20 provides, so this fixture uses two gene-rich chromosomes)
+        - Homo_sapiens.GRCh38.111_chr19_22.pc.gtf.gz: chr19+chr22 of Ensembl 111 GTF, subset to `gene_biotype "protein_coding"` with lean attributes
+        - bams/SRX1178088{5,6,7,8}.chr19_22.ds50.bam(+ .bai): 4-sample Ribo-seq cohort (same upstream SRX accessions as the chr20 BAMs above) at 50% downsample, filtered to chr19+chr22 and to reads overlapping protein-coding gene loci. 4 samples is the empirical PRICE-cohort floor; 3 samples crashes its noise-model inference.
+        - README.md: full derivation recipe, empirical justification for the chosen subset/cohort, and the PRICE invocation used for verification.
       - ribocode
         - genome_updated.gtf.gz: GTF with gene names updated via ribocode/gtfupdate, compressed for efficient storage
         - annotation.tar.gz: Tarball containing annotation directory output from ribocode/prepare for testing ribocode/metaplots and ribocode/ribocode modules

data/genomics/homo_sapiens/riboseq_expression/price/README.md

@@ -0,0 +1,39 @@
+# Test data for `gedi/price`
+
+A minimal cohort of four Ribo-seq samples covering protein-coding regions of chr19 + chr22 (Ensembl GRCh38). Genome reference and BAMs are masked/filtered to the protein-coding loci so that PRICE's expectation-maximisation converges and produces a non-empty `orfs.tsv` while keeping every fixture file under 4 MiB and the whole set under 11 MB.
+
+## Why this minimal set?
+
+PRICE's ORF inference fails (`Index 0 out of bounds for length 0` in `PriceOrfInference`) when too few candidate ORFs feed the noise model. Empirically:
+
+- chr19 alone, six samples: PRICE estimates the model but returns zero ORFs.
+- chr19 + chr22, three samples: PRICE crashes in `PriceOrfInference`.
+- chr19 + chr22, four samples, protein-coding-only reference: 381 ORFs - works.
+
+Restricting the reference to protein-coding genes (drops ~10% of ORF calls vs the full annotation), filtering BAMs to protein-coding gene loci (drops reads PRICE cannot use anyway), and keeping the four deepest samples is the minimum that keeps the test meaningful.
+
+## Files
+
+| File | Size | Description |
+|---|---|---|
+| `Homo_sapiens.GRCh38_chr19_22.pc_exon_masked.fa.gz` | 3.3 MB | chr19+chr22 from Ensembl GRCh38 primary assembly; everything outside protein-coding-gene exons hard-masked to `N`. |
+| `Homo_sapiens.GRCh38.111_chr19_22.pc.gtf.gz` | 1.6 MB | chr19+chr22 from Ensembl 111 GTF, subset to `gene_biotype "protein_coding"`. Attribute column trimmed to `gene_id`, `transcript_id`, `gene_biotype`, `gene_name`, `transcript_biotype`. |
+| `bams/SRX1178088{5,6,7,8}.chr19_22.ds50.bam` | 1.0-1.5 MB each | Four Ribo-seq samples from GSE182201, STAR-aligned to GRCh38, filtered to chr19+chr22, downsampled to 50% with `samtools view -bs`, then filtered to reads overlapping protein-coding-gene intervals. |
+| `bams/*.bai` | <90 KB each | BAM indexes. |
+
+Total: ~11 MB across 10 files.
+
+## How they were derived
+
+1. Source Ribo-seq reads SRR15480788/9/90/91 (SRA accessions for SRX11780885-8, from GSE182201) were aligned to the Ensembl GRCh38 primary assembly + Ensembl 111 GTF using STAR (full-genome alignment, post-rRNA-filtering).
+2. Each genome-sorted BAM filtered to chr19+chr22 with `samtools view -bh -F 256 <bam> 19 22`.
+3. Downsampled to 50% with `samtools view -bh -s 42.50`.
+4. Filtered to reads overlapping protein-coding gene loci with `samtools view -bh -L pc_gene.bed`.
+5. FASTA built from `Homo_sapiens.GRCh38.dna.chromosome.{19,22}.fa.gz` (Ensembl 111), then N-masked outside the union of protein-coding-gene exon intervals.
+6. GTF subset to chr19+chr22 from `Homo_sapiens.GRCh38.111.chr.gtf.gz`, restricted to `gene_biotype "protein_coding"`, then stripped of non-essential attributes.
+
+## Verified
+
+Running `gedi -e IndexGenome` then `gedi -e Price` on this cohort yields a `run.orfs.tsv` with 381 lines (380 ORFs plus header).
+
+Used by `modules/nf-core/gedi/price/tests/main.nf.test`.