What does the HHblits '-B' parameter do?

Question

[David]

Hi,

Can somebody please put me out of my misery and let me know what the '-B' parameter in hhblits is supposed to be doing? Officially, it specifies the 'maximum number of alignments in alignment list'. So, I assumed that increasing the -B integer would increase the max. number of HMM/MSA database hits to my query HMM, and by extension the total number of sequences that would be included in my output MSA.

However, I've tried to systematically vary -B from 1 to 50, 75, 100, and 200 and the total number of sequences in the output MSA does not change. I also do not discern clear changes in the output summary file (.out). It does not seem to be having any effect at all?

Thank you in advance.

Answer 1

[Mensur Dlakic]

The number of alignments you are going to get depends on database size (can't get 500 alignments from a database that has 100 members) and the E-value threshold (all hits have E-values, and it is usually not desired to print those where E=1000000000). When the hits are above your set E-value threshold, they won't show up in alignments no matter how much you increase -B. Finally, -B is specifically for the number of alignments that will be reported. If you look at the maximum number of summary lines in the output (specified by -Z, also defaults to 500), you will only change the alignments with -B but hits will still be reported in the summary list.

[David]

Thank you very much for the reply. It makes sense but there is something that I am still confused by; I will go into the specifics.

I am using an E-value threshold of 1e-10 and there are 137 MSA/HMMs that pass this threshold and from this I get 1159 sequences in my final output MSA. I therefore assumed that setting -B to 100 would restrict the number of MSA/HMMs (from 137 to 100) that would contribute to the number of sequences in my final output MSA. In every case I still get 1159 final sequences.

You say that it specifies the number of alignments that will be reported but what does this mean exactly if it is not affecting the summary list in the .out file? Some other visual output that is not affecting my final result MSA file? Thank you again.

[Mensur Dlakic]

I think both -B and -Z affect only the visual side, meaning only what is in the report. It should not affect how many sequences you get in the output file.

If you want fewer sequences, I suggest you go with hhfilter and play with the diversity switch (-neff).

[David]

Thanks. That was my impression also but I just wanted to make sure before starting my analysis. That makes sense.