LocalColabFold

1 Overview | 2 Using | 2.1 Multicore and GPU capabilities

Overview

LocalColabFold: LocalColabFold is an installer script designed to make ColabFold functionality available on users' local machines.
- An example on

Using

Use the module name localcolabfold to discover versions available and to load the application.

A good example on using this application can be found at the Harvard Medical School: Using (Local)ColabFold on O2.

Please note that the module loading and Slurm script examples will need to be modified for our clusters.

Multicore and GPU capabilities

The localcolabfold application has a number of commands which can be used across multiple cores as well as use GPUs. Look at the various commands and their help. For example:

[]$ colabfold_batch -h
...
  --use-gpu-relax       Run OpenMM/Amber on GPU instead of CPU. This can significantly speed up the relaxation runtime, however, might lead to compatibility issues with CUDA.
                        Unsupported on AMD/ROCM and Apple Silicon. (default: False)

[]$ colabfold_search -h
...
  --threads THREADS     Number of threads to use. (default: 64)