Adding dataset for testing glm part

.gitignore

@@ -0,0 +1 @@
+.nf-test.log

README.md

@@ -1,39 +1,40 @@
-# ![nfcore/test-datasets](docs/images/test-datasets_logo.png)
+# nf-core/dartseq Test Dataset
 
-Test data to be used for automated testing with the nf-core pipelines
+## Overview
 
-> ⚠️ **Do not merge your test data to `master`! Each pipeline has a dedicated branch (and a special one for modules)**
+This directory contains test data for the nf-core/dartseq pipeline, which detects and quantifies m6A RNA modifications from DART-seq data by identifying APOBEC1-YTH-induced C-to-U editing events adjacent to modified adenosines.
 
-## Introduction
+Two kinds of test data are provided:
 
-nf-core is a collection of high quality Nextflow pipelines. This repository contains various files for CI and unit testing of nf-core pipelines and infrastructure.
+- A small, fully synthetic 4-sample dataset (`fastq/`, `samplesheet_4samples.csv`, `reference/`) used by the pipeline's default `-profile test`, covering the full alignment + Bullseye site-calling workflow end-to-end.
+- A subsetted real-data fixture (`glm/`) derived from an actual DART-seq experiment, used specifically to exercise the `BULLSEYE_R_GLM` differential-editing statistics step with real coverage/mutation counts.
 
-The principle for nf-core test data is as small as possible, as large as necessary. Please see the [guidelines](https://nf-co.re/docs/contributing/test_data_guidelines) for more detailed information. Always ask for guidance on the [nf-core slack](https://nf-co.re/join) before adding new test data.
+## Files
 
-## Documentation
+### Synthetic 4-sample dataset
 
-nf-core/test-datasets comes with documentation in the `docs/` directory:
+- `fastq/{A03,A04,C03,E01}_R{1,2}.fastq.gz` - Paired-end DART-seq reads for 4 samples (25,000 read pairs each). A03/C03 are `control` group, A04/E01 are `dart` (edited) group.
+- `samplesheet_4samples.csv` - nf-core-format samplesheet (`sample,fastq_1,fastq_2,group`) referencing the fastqs above by their `raw.githubusercontent.com` URL.
+- `contrasts_test.csv` - Example custom contrasts file (`contrast_id,edited_group,control_group,mode,min_edit,max_edit,fold_threshold,min_sites`) for testing the `--bullseye_contrasts` option.
+- `reference/genome.fa` - Single-contig (`chr22_test`) reference genome, 180,001 bp, covering an ~11-gene region of human chr22 (DDT, DDTL, DERL3, GSTT2, GSTT2B, GSTTP2, LOC391322, MIF, MIF-AS1, SLC2A11, SMARCB1).
+- `reference/genes.gtf` - GTF annotation matching `genome.fa` (286 records, same gene set as above).
+- `reference/bullseye.test.refFlat` - refFlat-format annotation matching `genome.fa`/`genes.gtf`, used by Bullseye's `Find_edit_sites.pl`/`quantify_sites.pl` (23 transcripts).
 
-1.  [Add a new test dataset](https://github.com/nf-core/test-datasets/blob/master/docs/ADD_NEW_DATA.md)
-2.  [Use an existing test dataset](https://github.com/nf-core/test-datasets/blob/master/docs/USE_EXISTING_DATA.md)
+### GLM real-data fixture (`glm/`)
 
-## Downloading test data
+- `glm/coverage.txt`, `glm/mut.txt`, `glm/score.txt` - `BULLSEYE_GATHER_SITES` output matrices (per-site coverage / mutation count / score), restricted to `chr19` (3,143 sites). Derived from a single real DART-seq A04-vs-A03 contrast, reprocessed through the current pipeline's `BULLSEYE_QUANTIFY_SITES` + `BULLSEYE_ACFILTER` + `BULLSEYE_GATHER_SITES` modules to guarantee the file format matches what the pipeline currently produces.
+- `glm/observations.tsv` - Matching `BULLSEYE_R_GLM` design/colData file (`sample`, `group` columns).
 
-Due the large number of large files in this repository for each pipeline, we highly recommend cloning only the branches you would use.
+  Only one real sample/contrast is available in this fixture, so the beta-binomial GLM design has no residual degrees of freedom - `glm_comp_bbn()` in Bullseye's `functions.R` returns all-`NA` p-values by design whenever a comparison factor has fewer than 2 levels, rather than erroring. This fixture exercises `BULLSEYE_R_GLM`'s file-format handling and R execution path with real biological data; it is not a statistically meaningful GLM comparison (that would require multiple real contrasts, which this dataset does not have).
 
-```bash
-git clone <url> --single-branch --branch <pipeline/modules/branch_name>
-```
+**Data provenance and permission:** the `glm/` files derive from RNA-seq data generated for a METTL3 DART-seq study; redistributed here in subsetted, aggregated form (per-site coverage/mutation counts only - no raw reads or alignments) with permission from the data owner.
+
+## Usage
 
-To subsequently clone other branches[^1]
+Use the test profile in nf-core/dartseq:
 
 ```bash
-git remote set-branches --add origin [remote-branch]
-git fetch
+nextflow run nf-core/dartseq -profile test,docker
 ```
 
-## Support
-
-For further information or help, don't hesitate to get in touch on our [Slack organisation](https://nf-co.re/join/slack) (a tool for instant messaging).
-
-[^1]: From [stackoverflow](https://stackoverflow.com/a/60846265/11502856)
+The test configuration automatically references the samplesheet and reference files from this repository via `params.pipelines_testdata_base_path`.