1. Intent
Seqera
Pipeline: nf-core/rnaseq
Process: STAR_ALIGN
Profile: awsbatch
cpus = 64
memory = 488.GB
2. Execution (AWS Batch / Kernel)
Execution Reality
Nextflow task → Batch job
Instance: r5.16xlarge
Container: ..
Task: STAR_ALIGN (job 4521)
Duration: 47 min
Nextflow requested resources statically. Runtime behavior was not observable from logs or directives.
3. Reality
Tracer/
collect
Observed inside STAR_ALIGN container
nf-core/rnaseq → STAR_ALIGN → job 4521
CPU
8%
Memory
18%
Disk I/O
114%
92% CPU idle, 82% RAM idle. I/O bound. Adding CPU/RAM won't help.
4. The Fix
Tracer/
tune
Suggested for process STAR_ALIGN:
cpus = 8
memory = 64.GB
disk = local NVMe
Instance: r6id.2xlarge
This configuration will save ~85% on compute costs and improve runtime by 3.7x