Feature request: cluster size histogram

[joelb123]

Anybody who has to deal with clusters will need to make use of the size distribution at some point. It's not hard to cough up a script to take the cluster file and make it into a cluster size distribution histogram, but it would be better done once in mmseqs itself. Other useful columns for the cluster size distribution would be the percent of genes, the percent of clusters in that bin, and cumulatives on those starting from size 1 (singletons).

It would also be useful to calculate the following summary statistics and log them as well as writing them to a JSON file:

• Number of sequences in
• Total number of characters in the sequences
• Average sequence length
• Number of singletons
• Fraction of genes in singletons
• Size of largest cluster
• Fraction of genes in largest cluster
• Modal cluster size (peak of size distribution)
• Fraction of genes in modal cluster
• Which space was used (NA or AA)
• Run parameters
• processors used
• run times

Thanks for listening.

[milot-mirdita]

I like the idea, we have a summarizing module for taxonomy with taxonomyreport, doing something analogous for clustering sounds useful.