sokrypton / sokrypton/ColabFold

Discrepancies in Template Search Between Server and Local With Same Input MSA

Open
#676 0 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
Jupyter Notebook
Stars
2.9k
Forks
747
PR merge metrics
No merged PRs in 30d

Description

Expected Behavior

When running ColabFold locally, the template search should produce identical results to those generated by the server. Specifically, the templates returned from the local run should match those returned from the server, given that both use the same MSA and pdb70.m8 input files.

Current Behavior

There is a discrepancy in template search results. The server returns a different set of templates compared to the local run, despite using the same input MSA and pdb70.m8 files.

Steps to Reproduce

I obtained MSA and pdb70.m8 files from a server run.
Ran the local command:
colabfold_batch --templates --local-pdb-path /runtime/databases/mmseqs_2-colabfold/pdb/divided --pdb-hit-file ./server_generated_pdb70.m8 --num-seeds 5 --num-models 5 ./server_generated_msa.a3m ./output_folder

And got this:
2025-01-15 08:31:45,655 Sequence 0 found templates: ['8f7w_A', '8efb_A', '7xa3_A', '7x9y_A', '7vl8_A', '7vla_A', '7o7f_A', '6dde_A', '7sbf_A', '7u2l_A']
2025-01-15 08:31:54,201 Sequence 1 found templates: ['7vla_R', '7o7f_C', '7vl8_R', '6meo_B', '7x9y_R', '7f1q_R', '7xa3_R', '4n6h_A', '7t10_R', '7yit_A', '8ef5_M', '8ef5_R', '4rwa_B', '8efb_R', '4rwa_A', '8f7w_R', '7u2l_D', '7ul4_A', '6b73_A', '7sbf_R']

Compared to doing the template search on the server:
colabfold_batch --templates --num-seeds 5 --num-models 5 ./server_generated_msa.a3m ./output_folder

Which gives:
2025-01-14 06:42:50,723 Sequence 0 found no templates
2025-01-14 06:43:04,722 Sequence 1 found templates: ['7vla_R', '7vl8_R', '7o7f_C', '7xa3_R', '6meo_B', '7f1q_R', '7x9y_R', '4n6h_A', '4rwa_B', '8ef5_M', '8efb_R', '7t10_R', '7ul4_A', '7u2l_D', '7sbf_R', '6dde_R', '6vi4_A', '6b73_A', '8f7w_R', '7yit_A']

Given my input, I would expect the server’s results for the template search rather than the local ones.

Any help with understanding why this discrepancy is occurring will be much appreciated!
Thanks

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start by comparing the two shown colabfold_batch invocations and their server-generated_msa.a3m and server_generated_pdb70.m8 inputs. Trace the template-search entry point and compare the local and server template lists for both sequences; done means identifying why identical inputs produce different results or documenting the relevant environmental difference.

Written by the indexing model from the issue text.

Assessment

Tech stack
jupyter-notebook
Domain
bioinformatics
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Mostly clear
Newbie friendliness
32/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.