The effect of --min-seq-id parameter

[jianye00]

If I use cluster workflow, does the --min-seq-id parameter guarantee the alignment identity between the representative and cluster members to be at least what is specified by --min-seq-id ?

[milot-mirdita]

With the default cascaded clustering, this property can be broken due to transitivity issues with the repeated clustering. We have implemented an extended workflow that is enabled with --cluster-reassign that fixes these issues after the clustering is complete.

[jianye00]

Hi, I suppose you were answering my other issue #961 here about missed sequence after clusterupdate.

I tried --cluster-reassign flag too but it gives the same result as without it. Here is my command line:

bin/mmseqs clusterupdate --cluster-reassign 1 sequenceDB updateSequenceDB clusterDB newSequenceDB.cluster-reassign newClusterDB.cluster-reassign tmp

bin/mmseqs createtsv newSequenceDB.cluster-reassign newSequenceDB.cluster-reassign newClusterDB.cluster-reassign newClusterDB.cluster-reassign.tsv

wc newClusterDB.cluster-reassign.tsv
7313 14626 123018 newClusterDB.cluster-reassign.tsv
grep X7SER2 newClusterDB.cluster-reassign.tsv

As you can see, the cluster result using the "--cluster-reassign 1" flag generates 7313 clusters which is the same as without using it as shown in #961 and the entry X7SER2 is still missing.

[milot-mirdita]

We are currently looking at the other issue