RoseTTAFold All-Atom¶

Image Details
Running RoseTTAFold All-Atom Non-interactively
Available RoseTTAFold Versions

Compute Resources

Have questions or need help with compute, including activation or issues? Follow this link.
User Agreement

Docker Usage

The information on this page assumes that you have a knowledge base of using Docker to create images and push them to a repository for use. If you need to review that information, please see the links below.
Docker and the RIS Compute Service
Docker Basics: Building, Tagging, & Pushing A Custom Docker Image

storageN

The use of storageN within these documents indicates that any storage platform can be used.
Current available storage platforms:
- storage1
- storage2

Image Details ¶

Includes RoseTTAFold All-Atom

Docker Image¶

ghcr.io/washu-it-ris/rosettafold-all-atom-docker:<tag>

Docker Tag

The <tag> will refer to the version of the Docker container. Please click here to see a current list of available versions and their corresponding Docker images.

Running RoseTTAFold All-Atom Non-interactively ¶

Copy the script runscript.sh

#! /bin/bash

cd /app/RoseTTAFold/RoseTTAFold-All-Atom/
source /app/RoseTTAFold/mambaforge/condabin/activate RFAA
python -m rf2aa.run_inference --config-name $1

Copy the script bsub-non-interactive.sh

#! /bin/bash

# SPECIFY ACCORDINGLY
COMPUTE_GROUP=MY-COMPUTE-GROUP

# where are the fasta or sfd files among others?
INPUT_PATH=MY-INPUT-PATH

# where do you want the resulting files?
OUTPUT_PATH=MY-OUTPUT-PATH

# where are the config inference files?
INFERENCES_PATH=MY-CONFIG-INFERENCE-PATH

# Required by LSF
QUEUE=general
MEMORY=64GB
SLOTS=4

# The image tag should be a tag of the rosettafold-all-atom-docker
# or an image based on the rosettafold-all-atom-docker image.
IMAGE_TAG="ghcr.io/washu-it-ris/rosettafold-all-atom-docker:1.0.0"

# DO NOT CHANGE BELOW THIS LINE
ROSETTAFOLD_ASSETS_PATH="/scratch1/fs1/ris/references/RoseTTAFold/bfd:/app/RoseTTAFold/RoseTTAFold-All-Atom/bfd \
    /scratch1/fs1/ris/references/RoseTTAFold/checkpoint:/app/RoseTTAFold/RoseTTAFold-All-Atom/checkpoint \
    /scratch1/fs1/ris/references/RoseTTAFold/pdb100_2021Mar03:/app/RoseTTAFold/RoseTTAFold-All-Atom/pdb100_2021Mar03 \
    /scratch1/fs1/ris/references/RoseTTAFold/UniRef30_2020_06:/app/RoseTTAFold/RoseTTAFold-All-Atom/UniRef30_2020_06 \
    /scratch1/fs1/ris/references/RoseTTAFold/blast-2.2.26:/app/RoseTTAFold/RoseTTAFold-All-Atom/blast-2.2.26 \
    /scratch1/fs1/ris/references/RoseTTAFold/csblast-2.2.3:/app/RoseTTAFold/RoseTTAFold-All-Atom/csblast-2.2.3"

USER_CONFIGURED_PATH="$OUTPUT_PATH:/app/RoseTTAFold/RoseTTAFold-All-Atom/outputs \
    $INPUT_PATH:/app/RoseTTAFold/RoseTTAFold-All-Atom/inputs \
    $INFERENCES_PATH:/app/RoseTTAFold/RoseTTAFold-All-Atom/rf2aa/config/inference"

DOCKER_VOLUMES="$USER_CONFIGURED_PATH \
    $ROSETTAFOLD_ASSETS_PATH \
    $LSF_DOCKER_VOLUMES"

# BSUB
PATH="$PATH:/app/RoseTTAFold/mambaforge/condabin:/app/RoseTTAFold/mambaforge/bin" \
LSF_DOCKER_VOLUMES="$DOCKER_VOLUMES" \
bsub \
-n $SLOTS \
-M $MEMORY \
-R "rusage[mem=$MEMORY]" \
-R "gpuhost" \
-G $COMPUTE_GROUP \
-q $QUEUE \
-gpu 'num=1' \
-Is -a "docker($IMAGE_TAG)" /bin/bash

Keep both scripts in the same directory.
Change the variables such as the COMPUTE_GROUP, OUTPUT_PATH, INFERENCES_PATH among others.
E.g.

# SPECIFY ACCORDINGLY
COMPUTE_GROUP=compute-a.santiago

# where are the fasta or sfd files among others?
INPUT_PATH=$STORAGEN/rfaa/fastas/

# where do you want the resulting files?
OUTPUT_PATH=$STORAGEN/rfaa/outputs

# where are the config inference files?
INFERENCES_PATH=$HOME/rfaa/configs

# Required by LSF
QUEUE=qa
MEMORY=32GB
SLOTS=4

Copy the base.yaml config file and put it in the config/inference directory.

job_name: "structure_prediction"
output_path: outputs/
checkpoint_path: checkpoint/RFAA_paper_weights.pt
database_params:
  sequencedb: ""
  hhdb: "pdb100_2021Mar03/pdb100_2021Mar03"
  command: make_msa.sh
  num_cpus: 4
  mem: 64
protein_inputs: null
na_inputs: null
sm_inputs: null
covale_inputs:  null
residue_replacement: null

chem_params:
  use_phospate_frames_for_NA: True
  use_cif_ordering_for_trp: True

loader_params:
  n_templ: 4
  MAXLAT: 128
  MAXSEQ: 1024
  MAXCYCLE: 4
  BLACK_HOLE_INIT: False
  seqid: 150.0


legacy_model_param:
  n_extra_block: 4
  n_main_block: 32
  n_ref_block: 4
  n_finetune_block: 0
  d_msa: 256
  d_msa_full: 64
  d_pair: 192
  d_templ: 64
  n_head_msa: 8
  n_head_pair: 6
  n_head_templ: 4
  d_hidden_templ: 64
  p_drop: 0.0
  use_chiral_l1: True
  use_lj_l1: True
  use_atom_frames: True
  recycling_type: "all"
  use_same_chain: True
  lj_lin: 0.75
  SE3_param:
    num_layers: 1
    num_channels: 32
    num_degrees: 2
    l0_in_features: 64
    l0_out_features: 64
    l1_in_features: 3
    l1_out_features: 2
    num_edge_features: 64
    n_heads: 4
    div: 4
  SE3_ref_param:
    num_layers: 2
    num_channels: 32
    num_degrees: 2
    l0_in_features: 64
    l0_out_features: 64
    l1_in_features: 3
    l1_out_features: 2
    num_edge_features: 64
    n_heads: 4
    div: 4

Create a config inference yaml file.
- The inference config file should inherit the base settings from the base.yaml file.
- The defaults should be base and _self_.
- Set the job name; otherwise, the default job name will be “structure_prediction.”
- All the input files should have be prefixed with inputs/.
- For example, the K7N608.fasta file should be set as inputs/K7N608.fasta
- Example config file named Vmn1r78_E1050.rtaa.yaml

defaults:
  - base
  - _self_

job_name: "K7N608"

protein_inputs:
  A:
    fasta_file: inputs/K7N608.fasta

sm_inputs:
  B:
    input: inputs/66430.sdf
    input_type: "sdf"

Launch the non-interactive app

# Usage ./bsub-non-interactive.sh <<inference config name>>
# Do not include the yaml file extention.

./bsub-non-interactive.sh Vmn1r78_E1050.rtaa

Go to the OUTPUT_PATH to find the results. A successful run should generate a pdb file.

drwx------+  Aug 26 11:16 2024-08-26
drwx------+  Aug 26 10:32 K7N608
-rw-------+  Aug 26 10:38 K7N608_aux.pt
-rw-------+  Aug 26 10:38 K7N608.pdb

The output log should look like the following

Predicting:   0%|          | 0/1 [00:00<?, ?sequences/s]
Predicting: 100%|â–ˆâ–ˆâ–ˆâ–ˆâ–ˆâ–ˆâ–ˆâ–ˆâ–ˆâ–ˆ| 1/1 [00:00<00:00,  1.49sequences/s]
Predicting: 100%|â–ˆâ–ˆâ–ˆâ–ˆâ–ˆâ–ˆâ–ˆâ–ˆâ–ˆâ–ˆ| 1/1 [00:00<00:00,  1.48sequences/s]
Running HHblits against UniRef30 with E-value cutoff 1e-10
- 20:47:40.116 INFO: Input file = outputs/K7N608/A/hhblits/t000_.1e-10.a3m

- 20:47:40.116 INFO: Output file = outputs/K7N608/A/hhblits/t000_.1e-10.id90cov75.a3m

- 20:47:40.240 WARNING: Maximum number 100000 of sequences exceeded in file outputs/K7N608/A/hhblits/t000_.1e-10.a3m

- 20:48:50.358 INFO: Input file = outputs/K7N608/A/hhblits/t000_.1e-10.a3m

- 20:48:50.359 INFO: Output file = outputs/K7N608/A/hhblits/t000_.1e-10.id90cov50.a3m

- 20:48:50.646 WARNING: Maximum number 100000 of sequences exceeded in file outputs/K7N608/A/hhblits/t000_.1e-10.a3m

Running PSIPRED
Running hhsearch
Using the cif atom ordering for TRP.
./make_msa.sh inputs/K7N608.fasta outputs/K7N608/A 4 64  pdb100_2021Mar03/pdb100_2021Mar03
Aug 26 15:51:16 docker1_starter[2469871]: DEBUG:       run_docker returns 0
Aug 26 15:51:16 docker1_starter[2469871]: DEBUG:       Cleaning up temp files
Aug 26 15:51:16 docker1_starter[2469871]: DEBUG:       About to exit 0

Available RoseTTAFold Versions ¶

Current Version:¶

1.0.0

RoseTTAFold All-Atom¶

Image Details ¶

Docker Image¶

Running RoseTTAFold All-Atom Non-interactively ¶

Available RoseTTAFold Versions ¶

Current Version:¶

Table of Contents

Related Topics

This Page

RoseTTAFold All-Atom¶

Image Details¶

Docker Image¶

Running RoseTTAFold All-Atom Non-interactively¶

Available RoseTTAFold Versions¶

Current Version:¶

Image Details ¶

Running RoseTTAFold All-Atom Non-interactively ¶

Available RoseTTAFold Versions ¶