read length restriction in QIIME

[kramadu]

Hello,

I have a question about whether it is advantageous to restrict read lengths for the V6 region that is expected to be 110-130 base pairs. At the present, I have it restricted so that any reads below 100 base pairs and above 135 base pairs are not written. This is after the paired ends are joined. The reference for this is attached. My PI said those sequences may still be useful but wanted to make sure with people that are more knowledgeable in the field. Thanks in advance!

[sarvidsson]

How much of your sequences are outside of the expected length spectrum? If your preprocessing and OTU picking pipeline properly handles contaminants, it shouldn't hurt too much to include a little bit of odd sized fragments, however do check that
- you've clipped all sequencing adapters
- you've clipped of the primers and checked the sequences deeply for primer dimers

Each 16S primer set has it own problems, so you may (generally speaking - we don't do short V6 amplicons so I'm not familiar with the typical problems there) get odd-sized products from eukaryotic 18S or whatever other stuff you have in the sample (especially highly diverse samples with low concentration can deliver weird PCR fragments).

I do not use QIIME to pick OTUs (I'm more comfortable with Mothur), so I'm not sure if it is standard procedure to check for contaminating taxa - but I'd recommend you to do so.