- Disable MSA to reduce GPU memory for large complexes
- Default to 1 diffusion sample to fit within GPU VRAM
- Add errorStrategy 'ignore' so pipeline completes even if
individual tasks fail (standard bioinformatics pattern)
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Previous runs left rcsb_pdb_1QGS.fasta in the work dir.
Using s3://omic/eureka/chai1/input/ ensures only our files are picked up.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Pin transformers<4.45 and use v3 tag to ensure K8s pulls
the corrected image with working EsmModel imports.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
K8s caches :latest tag. Using :v2 ensures the permission-fixed
image is pulled instead of the cached old one.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Nextflow pipeline using chai1 Docker image from Harbor
- S3-based input/output paths (s3://omic/eureka/chai-lab/)
- GPU-accelerated protein folding with MSA support
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
