distace file with multiple processors

mothur bugs
1

Hi,
I’ve noticed a problem during what I interpret as the reassembling of temp distance files when multiple processors are used in dist.seqs. It sometimes happens that two lines are combined. Sometimes, not always. This results in an error that blocks the further clustering, since names are no more in sync in the two files. The error is like this:

********************###########
Reading matrix: |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
AAError: Sequence ‘BC08.02724’ was not found in the names file, please correct

In effect none of my sequence have this name becuase I labelled them with sample name followed by progressive integers, e.g. BC08.[1…N] so no “BC08.0” may be present. It is a merge of a true label and something else, e.g. another label or part of a distance.

Thanks for your support.

2

What version of mothur are you using?

3

I’m using the last version 1.23.1, compiled form source on Ubuntu Server 11.10 on an IBM server with 2 six core Xeon processors and 32 Gb RAM. I usually run dist.calc on 22 parallel threads.

4

Did you compile it using MPI? Is the file generated by dist.seqs in column, lower triangle or square format?

5

no, MPI is not enabled. The output of dist.seq is in column format.

6

If you built from source, could you try running the executable version from http://www.mothur.org/wiki/Download_mothur, to see if you get the same error?

7

I have a similar problem to report. The error report is …
AAError: Sequence ‘8’ was not found in the names file, please correct
after which the mothur exits. There should’nt be a way that any of my filenames in any of my files were “?8” to start with. So I am thoroughly confused.