From 5f1f639af3ddad8b5c4980f283b5ce9507429cf6 Mon Sep 17 00:00:00 2001 From: Fred Kolling Date: Wed, 13 May 2026 07:40:11 -0400 Subject: [PATCH] Increase picard_markdup maxtime from 30 min to 4 h The picard_markdup rule has maxtime="30:00", which is insufficient for larger BAMs. In a 12-sample mouse run (paired, HISAT2, ~5-6 GB sorted BAMs), 3 of 12 samples hit the 30-min wall and were killed mid-write while MarkDuplicates was still streaming records. Because cluster_profile/config.yaml does not define a status script, the Snakemake driver did not detect the SLURM TIMEOUTs and never exercised restart-times: 5 - the pipeline silently hung at 89% with no log activity until manually investigated. Bumping to 4:00:00 covers BAMs we've observed in practice (successful runs took 19-28 min) with comfortable headroom; users can still tune lower for small datasets via the resources override. Co-Authored-By: Claude Opus 4.7 (1M context) --- Snakefile | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/Snakefile b/Snakefile index 6375cfa..e82ec41 100644 --- a/Snakefile +++ b/Snakefile @@ -290,7 +290,7 @@ rule picard_markdup: conda: "env_config/picard.yaml", - resources: cpus="2", maxtime="30:00", mem_mb="20gb", + resources: cpus="2", maxtime="4:00:00", mem_mb="20gb", shell: """ {params.picard} -Xmx2G -Xms2G \