OneScience-Group/ProteinMPNN
124
1#!/bin/bash2 3 4SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"5PROJECT_ROOT="$(cd "${SCRIPT_DIR}/../.." && pwd)"6ONESCIENCE_ROOT="${ONESCIENCE_ROOT:-$(cd "${PROJECT_ROOT}/.." && pwd)}"7export PYTHONPATH="${PROJECT_ROOT}/model:${ONESCIENCE_ROOT}/src:${PYTHONPATH:-}"8 9 10#new_probabilities_using_PSSM = (1-pssm_multi*pssm_coef_gathered[:,None])*probs + pssm_multi*pssm_coef_gathered[:,None]*pssm_bias_gathered 11#probs - predictions from MPNN12#pssm_bias_gathered - input PSSM bias (needs to be a probability distribution)13#pssm_multi - a number between 0.0 (no bias) and 1.0 (no MPNN) inputed via flag --pssm_multi; this is a global number equally applied to all the residues14#pssm_coef_gathered - a number between 0.0 (no bias) and 1.0 (no MPNN) inputed via helper_scripts/make_pssm_input_dict.py can be adjusted per residue level; i.e only apply PSSM bias to specific residues; or chains15 16 17 18pssm_input_path="${PROJECT_ROOT}/data/inputs/PSSM_inputs"19folder_with_pdbs="${PROJECT_ROOT}/data/inputs/PDB_complexes/pdbs/"20 21output_dir="${PROJECT_ROOT}/outputs/example_pssm_outputs"22if [ ! -d "$output_dir" ]23then24 mkdir -p "$output_dir"25fi26 27path_for_parsed_chains=$output_dir"/parsed_pdbs.jsonl"28path_for_assigned_chains=$output_dir"/assigned_pdbs.jsonl"29pssm=$output_dir"/pssm.jsonl"30chains_to_design="A B"31 32python "${PROJECT_ROOT}/scripts/helper_scripts/parse_multiple_chains.py" --input_path="$folder_with_pdbs" --output_path="$path_for_parsed_chains"33 34python "${PROJECT_ROOT}/scripts/helper_scripts/assign_fixed_chains.py" --input_path="$path_for_parsed_chains" --output_path="$path_for_assigned_chains" --chain_list "$chains_to_design"35 36python "${PROJECT_ROOT}/scripts/helper_scripts/make_pssm_input_dict.py" --jsonl_input_path="$path_for_parsed_chains" --PSSM_input_path="$pssm_input_path" --output_path="$pssm"37 38python "${PROJECT_ROOT}/scripts/inference.py" \39 --jsonl_path "$path_for_parsed_chains" \40 --chain_id_jsonl "$path_for_assigned_chains" \41 --out_folder "$output_dir" \42 --num_seq_per_target 2 \43 --sampling_temp "0.1" \44 --seed 37 \45 --batch_size 1 \46 --pssm_jsonl "$pssm" \47 --pssm_multi 0.3 \48 --pssm_bias_flag 1 \49 --path_to_model_weights "${PROJECT_ROOT}/weight/vanilla_model_weights"50 