I am trying to get the lukasa workflow to work in Arvados. I have hit a few problems:
The workflow uses format annotation extensively. There is some sample input here in FASTA format. When I upload that to Arvados I cannot see a way to associate format info with the files.
The workflow fails at the metaeuk step, as far as I can tell because it is running with the default memory limit of 1GB RAM. I have tried to increase the memory limit through a ResourceRequirement (see below) but it does not lead to more RAM being allocated. Here is the start of the adapted workflow:
#!/usr/bin/env cwl-runner cwlVersion: v1.0 class: Workflow inputs: contigs_fasta: type: File proteins_fasta: type: File species_table: type: string? outputs: spaln_out: type: File outputSource: process_spaln_output/combined_spaln_output requirements: ResourceRequirement: ramMin: 4096 ramMax: 4096 coresMin: 1 coresMax: 4
The environment I am running this on is a VM with 16 GB RAM, 4 vCPUs running Ubuntu 18.04. Arvados was installed using the single-host Salt method.