Classify.seqs on shotgun type data

Hi All,

I have some metagenomic data from a shotgun type of experiment, not amplicons as mothur is usually used for. I would like to try to classify the 16S portion of that data with the mothur implementation of the RDP algorithm.
I figured it would work well because the algorithm is k-mer based, and there is no alignments of this data. As it is generated from random fragments of DNA it might also come from any strand.
When i classify the sequences i get a warning that approximately half the sequences is reversed (see below).

Is this something i should be concerned about, or is this algorithm not as well-suited for this type of data as i thought.

Thanks in advance


[WARNING]: mothur suspects some of your sequences may be reversed, please check batch1_1.TCA.454Reads.fna.coord_Bacterial_16S_rRNA_extract.sff_clean.pds.flip.accnos for the list of the sequences.

It took 1863 secs to classify 13456 sequences.

You can have classify.seqs flip the sequences automatically. I would expect half to be backwards since the 16S genes are probably randomly oriented in your dataset.

Thank you for the answer.
Do i have to set an option to classify.seqs or does it do it by default?
I do not see mention of this on the manual wiki.

Mothur automatically checks for reversed sequences when you use the wang method. It uses the template probabilities to determine if your sequence or the reverse compliment of your sequence will have a stronger match in the template. If the reverse is a better match then, we classify using the reverse and warn you.

Thank you for the reply, this sounds like what i was hoping for.