Bidirectional sequencing

Can Mothur handle 454 bidirectional sequencing? Would the oligos file include the forward primer, reverse primer, and the reverse complements of each? And then what about the barcodes?
Thanks very much,
Jeff

Yup, the side you sequenced from is forward. trim.seqs will look for the barcode and then the forward primer sequence.

Thanks, that’s very clear. What about barcodes following the reverse primer, or combinations of barcodes on the forward and reverse ends?

Nope, at this point the only place that we can handle barcodes is before the forward primer. Typically people are using 454 primer combinations so that they never really get to the reverse primer so it’s generally not used in trim.seqs.

I was actually wondering how to use mothur for bidirectional sequencing myself. I have a 27f/519r primer combination spanning V1 to V3. I know that

But in my application I was thinking to use the partial overlap between the forward and reverse amplicons to 1)increase confidence on the overlapping part and 2) have a ‘longer’ read length to increase certainty of classification, and if possible allow for more ‘in depth’ classification. There are some suggestions on how to do this with Qiime (https://groups.google.com/forum/?fromgroups=#!topic/qiime-forum/MJNNsXTQpUA), however I would rather use mothur as I already have a sweave-based report generating file based upon mothur’s output file formats, and I like to work with mothur better than with qiime.

So any suggestions on how to use mothur in this case are very much welcome. The data has already been run, without asking me if bi-directional was a good idea. However if I can only use the forward I am more or less throwing half of the data away.

ps: one nasty detail: I got the sff files already seperated per MID so I need to use sff.multiple in stead of the normal SOP. :oops:

This is from a post 1.5 yrs ago - a lot has happened since then :slight_smile:

Are you using 454? If so, then I think making contigs would set you up for a disaster with creating contigs since the same fragment is not sequenced in both directions.

Are you using Illumina? If so, then please see the make.contigs page.

Pat

Dear Pat,

Thanks for the reply. I am using 454, however rather then making ‘contigs’ I want to use the overlapping region to actually mainly create a longer ‘read’ to increase classification certainty. I am basing myself on this Plos One paper: http://www.plosone.org/article/info%3Adoi%2F10.1371%2Fjournal.pone.0040467#s2
Where according to them

Read affiliation was done for combined forward and reverse reads for each library using the RDP classifier

Here they apparently use the SEQMAN II software by DNAStar, after which a reference tree was made in ARB and the ARB parsimony tool was used to add the contigs to this tree, and like this the contigs were identified. (a more detailled workflow can be found in http://onlinelibrary.wiley.com/doi/10.1111/j.1574-6941.2011.01083.x/pdf).

I would like to do a likewise thing using mothur. For now I am trying (and failing) to write a perl/python implementation to make contigs because I do not have the SEQMAN II software, and my PI is not going to buy it for me (I am working together with one of the authors of the plos one paper and apparently right now they are using Lasergene 10, which will also not be bought). Computational power is not an issue. I have a 24 core 32GB ubuntu server all to myself and if I want I can get time on the university’s HPC.
I loathe working with ARB, so a mothur-style solution would be much appreciated.

Kind regards,

Frederiek-Maarten Kerckhof