View Single Post
Old 10-16-2015, 01:15 PM   #7
Senior Member
Location: CT

Join Date: Apr 2015
Posts: 237

If you are sequencing amplicons for 16s you need to cluster the sequences into OTUs. the more sequencing errors you have the more spurious OTUs you generate-which massively increases the memory require to cluster those (assuming you are doing de novo clustering). If you get them clustered, you then will waste a lot of time trying to find meaning in the sequencing noise or you can just throw out all of the rare OTUs which means that you will be throwing out good data along with the bad because you can't tell the difference between the good and bad rares. Ecologically this matters, for a cancer panel-maybe it doesn't.
thermophile is offline   Reply With Quote