OneScience-Group/AlphaFold3
3254
1#!/bin/bash2set -euo pipefail3 4SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"5PROJECT_ROOT="$(cd "${SCRIPT_DIR}/.." && pwd)"6 7 8export PATH="${PROJECT_ROOT}/flax_model/alphafold3/_tools/hmmer/bin:${PATH}"9export TF_CPP_MIN_LOG_LEVEL="${TF_CPP_MIN_LOG_LEVEL:-0}"10export JAX_TRACEBACK_FILTERING="${JAX_TRACEBACK_FILTERING:-off}"11export XLA_CLIENT_MEM_FRACTION="${XLA_CLIENT_MEM_FRACTION:-0.95}"12export TRITON_ENABLE_GLOBAL_TO_LOCAL="${TRITON_ENABLE_GLOBAL_TO_LOCAL:-1}"13export TRITON_USE_MAKE_BLOCK_PTR="${TRITON_USE_MAKE_BLOCK_PTR:-1}"14export TRITON_DEFAULT_ENABLE_NUM_VGPRS512="${TRITON_DEFAULT_ENABLE_NUM_VGPRS512:-1}"15export HIP_VISIBLE_DEVICES="${HIP_VISIBLE_DEVICES:-0}"16 17MODEL_ROOT="${ONESCIENCE_MODELS_DIR:-${PROJECT_ROOT}/weight}"18MODEL_DIR="${ALPHAFOLD3_MODEL_DIR:-${MODEL_ROOT}}"19DATASET_ROOT="${ALPHAFOLD3_DATASET_ROOT:-${ONESCIENCE_DATASETS_DIR:-${PROJECT_ROOT}/data}/alphafold3}"20DB_DIRS="${ALPHAFOLD3_DB_DIR:-${DATASET_ROOT}/public_databases}"21MMSEQS_DB_DIRS="${ALPHAFOLD3_MMSEQS_DB_DIR:-${DATASET_ROOT}/mmseqsDB}"22JSON_PATH="${ALPHAFOLD3_JSON_PATH:-${PROJECT_ROOT}/inputs/t1119_search.json}"23OUTPUT_DIR="${ALPHAFOLD3_OUTPUT_DIR:-${PROJECT_ROOT}/outputs}"24RUN_INFERENCE="${ALPHAFOLD3_RUN_INFERENCE:-false}"25FLASH_ATTENTION="${ALPHAFOLD3_FLASH_ATTENTION:-cutlass}"26 27mkdir -p "${OUTPUT_DIR}"28 29python "${PROJECT_ROOT}/scripts/run_alphafold.py" \30 --json_path="${JSON_PATH}" \31 --model_dir="${MODEL_DIR}" \32 --output_dir="${OUTPUT_DIR}" \33 --run_data_pipeline=true \34 --run_inference="${RUN_INFERENCE}" \35 --flash_attention_implementation="${FLASH_ATTENTION}" \36 --use_mmseqs=false \37 --use_mmseqs_gpu=false \38 --db_dir="${DB_DIRS}" \39 --mmseqs_db_dir="${MMSEQS_DB_DIRS}" \40 --small_bfd_database_path="${DATASET_ROOT}/jackhmmer_split/bfd-first_non_consensus_sequences.fasta@64" \41 --small_bfd_z_value=65928866 \42 --mgnify_database_path="${DATASET_ROOT}/jackhmmer_split/mgy_clusters_2022_05.fa@512" \43 --mgnify_z_value=623796864 \44 --uniprot_cluster_annot_database_path="${DATASET_ROOT}/jackhmmer_split/uniprot_cluster_annot_2021_04.fa@256" \45 --uniprot_cluster_annot_z_value=225619586 \46 --uniref90_database_path="${DATASET_ROOT}/jackhmmer_split/uniref90_2022_05.fa@128" \47 --uniref90_z_value=153742194 \48 --jackhmmer_n_cpu="${ALPHAFOLD3_JACKHMMER_N_CPU:-2}" \49 --jackhmmer_max_parallel_shards="${ALPHAFOLD3_JACKHMMER_MAX_PARALLEL_SHARDS:-1}" \50 --jackhmmer_max_threads="${ALPHAFOLD3_JACKHMMER_MAX_THREADS:-8}" \51 --nhmmer_n_cpu="${ALPHAFOLD3_NHMMER_N_CPU:-2}" \52 --nhmmer_max_parallel_shards="${ALPHAFOLD3_NHMMER_MAX_PARALLEL_SHARDS:-1}" \53 --nhmmer_max_threads="${ALPHAFOLD3_NHMMER_MAX_THREADS:-8}"54 