Hi,
I’m using OpenDDE with the default MSA generation, which internally uses colabfold_search and the public ColabFold/MMseqs2 API.
I’m seeing a recurring issue where a substantial fraction of the generated A3M files contain only the query sequence and no additional homologs. No error is reported by colabfold_search or OpenDDE—the job completes successfully, but the resulting MSA consists solely of the input sequence.
The strange part is that this appears to be random. If I rerun the exact same sequence, it often returns a normal MSA with many homologs. In other words, the behavior is not reproducible for a given sequence:
• Run 1 → query-only A3M
• Run 2 (same sequence) → normal MSA with many hits
Because the same sequence can produce completely different results across runs, I suspect this is not caused by the sequence itself but rather by the public API. I find the following error under this link:
https://search.mmseqs.com/
Is this the reason for my error? Do you know approx when it will be fixed?
Thanks!
Hi,
I’m using OpenDDE with the default MSA generation, which internally uses colabfold_search and the public ColabFold/MMseqs2 API.
I’m seeing a recurring issue where a substantial fraction of the generated A3M files contain only the query sequence and no additional homologs. No error is reported by colabfold_search or OpenDDE—the job completes successfully, but the resulting MSA consists solely of the input sequence.
The strange part is that this appears to be random. If I rerun the exact same sequence, it often returns a normal MSA with many homologs. In other words, the behavior is not reproducible for a given sequence:
• Run 1 → query-only A3M
• Run 2 (same sequence) → normal MSA with many hits
Because the same sequence can produce completely different results across runs, I suspect this is not caused by the sequence itself but rather by the public API. I find the following error under this link:
https://search.mmseqs.com/
Is this the reason for my error? Do you know approx when it will be fixed?
Thanks!