implementing-new-modules

Create a directory under multiqc/modules/toolname with an __init__.py and a parser class extending BaseMultiqcModule, then register a search pattern in search_patterns.yaml and an entry point in pyproject.toml. Finish with unit tests, an integration test via pytest tests/test_modules_run.py, and lint checks.

Iterate over self.find_log_files("toolname"), parse each file into a per-sample data dict, and call add_data_source and add_software_version for every sample. After filtering with ignore_samples, raise ModuleNoSamplesFound if empty, then add sections and call write_data_file last.

Use a multi-subtool structure when the tool has distinct subcommands emitting different output formats, like samtools or picard. Each subtool gets its own parser file with a parse_toolname_subtool function, coordinated by an orchestrator class that raises ModuleNoSamplesFound only if all subtools return zero samples.

Common causes are missing add_software_version calls, calling write_data_file before all sections are added, or raising UserWarning instead of ModuleNoSamplesFound. Run python .github/workflows/code_checks.py plus ruff and mypy to identify the exact violation.

Add an entry to search_patterns.yaml using fn for standard filenames, contents for exact string matches, or contents_re for regexes, with num_lines to cap scanning. Audit the tool's upstream source for version and flag-dependent output variations before finalizing the pattern.