AF Cluster

PaperCodeProduction Ready

Predicting multiple conformations via sequence clustering

Readiness

Readiness: Production Ready
Smoke: passed
Artifacts: passed
Production: production_ready_cpu_msa_clustering
Storage: <5
Strategy: shared python312_slim.sif + tool ext3 overlay
Next: Validate user-uploaded A3M/FASTA input from the web UI and normalize the curated subset into a clean shared package/archive.

SOP validation used upstream AF_Cluster revision 6b22451c5a9def576f43824bb5573a9c1faf6c55. Curated runnable subset is /media/nik/seagate_nik/bio_server/tools/af-cluster/AF_Cluster_subset (2.5M) with env overlay /home/nik/bio_mega_envs/af-cluster-py312-target (82M). Upstream full clone is 2.7G and should be replaced by a clean sparse/subset package later. Worker Slurm job 350 and API Slurm job 351 completed on CPU using tiny_two_family.a3m. API artifacts include TINY_clustering_assignments.tsv, TINY_cluster_metadata.tsv, clustered/control A3M files, results.csv, summary.json, stdout/stderr, and Slurm logs. Compatibility patch: ClusterMSA.py --min_samples parsed as int for current scikit-learn.