Flye 2.9.6+galaxy0 RAM allocation for a 150 Mb Arabidopsis HiFi pilot (compressed input)

usegalaxy.eu support
1

Hello Galaxy Europe Support,

Before submitting a small public-data assembly pilot on usegalaxy.eu, could you please confirm the appropriate resource allocation for:

toolshed.g2.bx.psu.edu/repos/bgruening/flye/flye/2.9.6+galaxy0

The proposed input is a uniform random subset of 250,000 PacBio HiFi reads from ENA ERR8666127 (Arabidopsis thaliana Col-0), sampled with seqtk seed 4, two-pass mode. Based on run metadata, the subset is expected to contain approximately 4.514 Gb of sequence, around 30.1x nominal coverage using a 150 Mb planning genome size. Actual subset bases and compressed size have not yet been measured. Flye would use --pacbio-hifi, one polishing iteration, without meta or scaffolding options. No Flye job has been submitted or failed; this is a resource-planning question.

The public shared TPV rule matching this tool prefix specifies 20 cores and:

mem: min(max(input_size * 1.2, 3.8), 256)

Our reading of legacy input_size is the sum of unique input datasets’ physical stored bytes divided by 2^30. Assuming compression similar to the submitted Col-0.q20.fastq.gz file, the planned subset would be approximately 3.628 GiB compressed, resulting in only approximately 4.35 GiB RAM under this rule. This is a static estimate, not a measured allocation. The compression-aware get_input_size helper appears separate from the legacy input_size used in this expression. We cannot verify the live configuration, applicable overrides or final destination.

Could you please:

  1. Confirm the currently deployed rule and how FASTQ.gz input_size is evaluated for this exact Flye version, including any applicable destination overrides.
  2. Confirm whether approximately 4.35 GiB would be appropriate for this approximately 150 Mb HiFi assembly; if not, advise the supported way to request a suitable allocation before submitting.
  3. Clarify the expected CPU/RAM/walltime and OOM resubmission policy for this pilot, and whether any relevant user resource controls are available.

We will keep Flye on hold pending clarification. We will not misdeclare the datatype/input_size, modify the wrapper or server configuration, or run jobs simply to probe memory. No sequence files, credentials or private research data are attached.

Public configuration references reviewed:

Thank you for advising a safe, supported allocation.