alphafold3

AlphaFold3 Example

In order to run AlphaFold3 with SBGrid you must: 1. Obtain the AF3 models 2. Download the AF3 databases

Obtain the AF3 models.

SBGrid cannot distribute these. You must apply through this form:

https://forms.gle/svvpY4u2jsHEwWYS6

The link for the form is also given in the git repository

You will receive a link to download the models as a zstd compressed file. Uncompress this file to produce af3.bin. Place this file in a directory of your choice. You must supply this directory as an argument to the run_alphafold.py script.

Download databases

The databases for AlphaFold3 (AF3) must be downloaded.

AlphaFold provides a script for downloading these in the github repo:

https://github.com/google-deepmind/alphafold3/blob/main/fetch_databases.sh

The version of AF3 curated in SBGrid uses the tar file version of the mmcif files in pdb_2022_09_28_mmcif_files.tar.

Since then AF3 has changed to use the uncompressed files in the mmcif tar file. The download script will untar the file.

The resulting layout should look like this:

/programs//local/alphafold-3.0.0/databases/
├── bfd-first_non_consensus_sequences.fasta
├── compressed
│   ├── bfd-first_non_consensus_sequences.fasta.zst
│   ├── mgy_clusters_2022_05.fa.zst
│   ├── nt_rna_2023_02_23_clust_seq_id_90_cov_80_rep_seq.fasta.zst
│   ├── pdb_2022_09_28_mmcif_files.tar.zst
│   ├── pdb_seqres_2022_09_28.fasta.zst
│   ├── rfam_14_9_clust_seq_id_90_cov_80_rep_seq.fasta.zst
│   ├── rnacentral_active_seq_id_90_cov_80_linclust.fasta.zst
│   ├── uniprot_all_2021_04.fa.zst
│   └── uniref90_2022_05.fa.zst
├── mgy_clusters_2022_05.fa
├── nt_rna_2023_02_23_clust_seq_id_90_cov_80_rep_seq.fasta
├── pdb_2022_09_28_mmcif_files.tar
├── pdb_seqres_2022_09_28.fasta
├── rfam_14_9_clust_seq_id_90_cov_80_rep_seq.fasta
├── rnacentral_active_seq_id_90_cov_80_linclust.fasta
├── uniprot_all_2021_04.fa
└── uniref90_2022_05.fa