Refactor active Snakefiles to Snakemake best practices

[caleblareau]

Follow-up to #42. Refactors the three Snakefiles driving the active commands (Snakefile.bap2.chr, Snakefile.bap_frags, Snakefile.bap_bulk_frags) from the run: + os.system(<string>) anti-pattern to idiomatic Snakemake.

Changes

• shell: directives replace string-concatenated os.system() calls, using {input}/{output}/{params}/{threads} and global variables.
• log: directives replace manual 2> / &> redirects.
• params: hold computed values (name-sorted bam path, temp/annotated paths, cut columns, QC flag toggles). awk programs are kept as readable named constants and substituted verbatim via {params.*} (avoids brace-escaping noise).
• determine_HQ_beads keeps run: (it has genuine Python: Counter/knee-calling logic) but swaps os.system() for Snakemake's shell() helper.
• Replaced pysam.sort/pysam.index with the configured samtools binary so those rules become clean shell: blocks, and fixed a latent NameError (undefined pp when --all-pairs is set) via a proper_pair_filter param.

Behavior preservation

Command semantics are preserved exactly. Every rule's rendered shell command was verified with snakemake -n -p against a scaffolded config + DAG (all three build valid DAGs: 17 / 5 / 4 jobs). Subtle differences were kept intact — e.g. bap_frags gzips its bedpe while bap_bulk_frags does not, the nc_threshold argument position, samtools collate vs sort -n, and the tabix <file> -p bed argument order. One real bug was caught and fixed during validation: a bare {name} collided with Snakemake's reserved shell-namespace key and rendered the rule name; it now passes the sample name via a sample_name param.

Out of scope

The deprecated v1 Snakefile.bap.chr is left unchanged, consistent with leaving the v1 CLI untouched in #42.

🤖 Generated with Claude Code