Hey all..I've been testing the latest colabfold version on the dev branch with the multimer_v3 model. It seems to run OK when downloading the DBs and then resuming with this command:

NXF_VER=25.10.4 nextflow run proteinfold -profile singularity,gpu --input /path/to/proteinfold/assets/samplesheet.csv --outdir output --use_gpu true --mode colabfold --colabfold_model_preset alphafold2_multimer_v3 --colabfold_use_amber false

colabfold_use_amber false was required else there was an error:

ValueError: Minimization failed after 100 attempts.

Similarly the memory was not enough so had to change the memory for this:

withName: 'MMSEQS_COLABFOLDSEARCH' {
        memory = { 400.GB   * task.attempt }