00. Overview and decision points¶
Tutorial contents · Next: Resource planning →
Objective¶
Convert paired-end bulk ATAC-seq reads into a quality-controlled region-by- sample fragment-count matrix, test a prespecified accessibility contrast, and report the result with appropriate multiple-testing control and limitations.
Decisions that must not be copied blindly¶
Before running commands, decide and record:
- organism, genome assembly, chromosome naming convention, and reference release;
- library layout, adapter sequences, and whether mitochondrial or alternate contigs are retained;
- biological unit, replicate structure, batches, paired donors, covariates, reference level, and contrast;
- mapping-quality, duplicate, blacklist, peak-calling, and low-count rules;
- whether peaks are narrow or broad and how the consensus universe is built;
- primary adjusted-p-value and effect-size thresholds;
- the universe used for enrichment and any explicitly exploratory analyses.
The internship used GRCh38, paired-end libraries, a donor-aware model, and a 10-count-in-5-samples filter because those choices matched that dataset and the agreed analysis plan. They are an example, not universal defaults.
Reproducibility rule¶
For every stage, preserve four things: immutable input identity, the exact command and software version, validation evidence, and a new non-overwriting output location. Separate observed results from interpretation.