chimera.seqs(method=bellerophon) left=right

[1.7.2,Win32]

Output from chimera.seqs(method=bellerophon) shows several examples where both the left and right parts of the chimera are coming from the same sequence! Example:

Name Score Left Right 
...
S_F094P_F09 1.72 G_A112I_A11 G_A112I_A11

Let me know if you would like me to e-mail my input file.

Robin

aaaand the perl 1-liner workaround:

perl -ne "print unless /(\w+)\t\1\r\l/" messedup_chimeras.txt > good_chimeras.txt

Robin

That’s a feature :). Could you send the sequence collection to mothur.bugs@gmail.com?

Thanks for the catch,
Pat

Sent.

In the handful of examples I examined manually, the putative chimera differed from the target by a single base. It also occurred to me that the chimera.seqs command won’t stop the user from running it on a file that hasn’t been processed with unique.seqs. That’s not the issue in my case, but I would certainly like to see mothur do more to keep the user from making an ass of herself. (Maybe check for the word “unique” in the filename and throw a warning if it isn’t there?) I guess I’m asking for better assproofing :wink:

Also, a minor suggestion: instead of reporting “Sequence with preference score above 1.0”
I would like to see the cutoff increased to 1.1. This is the Bellerophon server default and matches my own experience of staring (and staring) at putative chimera alignments.

Robin

If we ass-proofed mothur, then I probably couldn’t use my own software! :). I hear your concern, but if you couldn’t tell, we give people a lot of opportunities to make an ass of themselves. Take for instance the myriad of options available for chimera checking.

…but for a program called mothur, don’t you think a little more nagging is in order? :mrgreen:

Mothur, do you think she’s good enough…