We’ve encountered a really random bug. When running Mothur (v.1.20.3 or 1.21) on an older iMac (2006 Intel Core 2 Duo, 2 GB RAM, OS 10.5.8) align.seqs doesn’t work correctly. To be specific we loaded a data set that had ~150K sequences (as of the *.trim.unique.fasta file) and tried to align.seqs. The alignment appears to proceed normally, but the after running summary.seqs only 1 sequence is present (the alignment file is still huge (2.5 GB) though). We have tried this on a couple computers of the same model year and get the same result. I’ve run the data on my tower and it proceeds just fine. Just in case there was something goofy about the file I’ve used the *trim.unique.fasta file from my computer (which aligns correctly) and tried to align it on the iMacs and still no go. I’ve recompiled 1.20.3 from source on the iMac and still get the same result. Anyone else noticed this?
sigh I guess no one has. Well here’s an update anyway:
My original statement that align.seqs does not work is incorrect. It works on small datasets, like Costello, but does not seem to work on large data sets ~400k sequences. Oddly, the .align.report looks ok. But when you do summary.seqs, or try to continue, the file looks bad. It’s 16 GB in size, but seems to be 1 sequence. The summary.seqs produces the following result:
Before alignment
Start End NBases Ambigs Polymer
Minimum: 1 173 173 0 2
2.5%-tile: 1 218 218 0 4
25%-tile: 1 418 418 0 5
Median: 1 475 475 0 5
75%-tile: 1 498 498 0 5
97.5%-tile: 1 512 512 1 7
Maximum: 1 654 654 7 8
of unique seqs: 349407
total # of seqs: 419500
After alignment
Start End NBases Ambigs Polymer
Minimum: 1044 13862 490 0 5
2.5%-tile: 1044 13862 490 0 5
25%-tile: 1044 13862 490 0 5
Median: 1044 13862 490 0 5
75%-tile: 1044 13862 490 0 5
97.5%-tile: 1044 13862 490 0 5
Maximum: 1044 13862 490 0 5
of unique seqs: 1
total # of seqs: 1
Oh, and to expand it further it seems to affect a newer iMac as well. None of them have more than 2 GB of memory so perhaps it’s a memory allocation issue. Again, any thoughts would be appreciated.