I'm working with sequences of 16S (V1V2 region, around 311 bp) using IonTorrent sequencing. I run "demux summarize" command to get the plot to decide the truncation lenght. demux-single-end.qzv (283.8 KB)
I red in QIIME2 forum that for IonTorrent sequences they recommend to cut at the first position in which the Median drops down 25, is it correct? As I worked before with QIIME1, and using the split libraries command for truncating my sequences at 300, I run the denoise single-end with the same truncation lenght.
I made two tests with denoise-single command, in which my truncation lenght was 300 and 386 respectively; I obtained in the first test a minimum frequency per sample of 145,041 and a maximum of 502,289. In the second one I obtained a minimum frequency per sample of 2 and a maximum of 90.
So if my amplicon is around 311 bp, which would the correct lenght be, considering the quality plot, from demux command?
Can you provide a link to that thread? I couldn't find that recommendation after a quick search of the forum.
As far as sequence trimming and truncation recommendations, to the best of my knowledge there aren't any hard and fast "rules," it really comes down to investigator interpretation, as well as any needs specific to your study. We generally recommend that users explore a variety of settings for these parameters, and then determine their impact on some downstream steps (for example, how many features are present in you feature table?). One thing worth noting, the official DADA2 recommendation for IonTorrent data is to set the trim-left value to 15.
I was looking for the thread before posting my question but I have not found it yet, I will check again and post it if I will find it!
I was discussing about the median value with my collagues and as in QIIME1 the minimum quality score used was 25 in split libraries command, can we consider the same value in QIIME2?
I checked the section you suggest, but as in this section the trim left is used to remove primers (“you can also use the trimLeft argument of dada2’s filtering functions to remove the primer nucleotides”) and we did it using trim-left 19 (because it is the lenght of our Fw primer), would it be correct to set the trim-left at 15?
Disclaimer: I have not analyzed IonTorrent data before (in QIIME 2, or otherwise), so my suggestions are based on my interpretation of @benjjneb’s documentation. I think besides trimming your primers, you will also need to trim an additional 15 nt, so --p-trim-left 34 (which is 19+15), but again, I can’t comment specifically on if this is correct or not, this is just what I would do. Personally, I would experiment tweaking the trim and trunc parameters and see how that impacts your results. Thanks!
In this case, if trimming off the first 19nts, no need to trim another 15nts. Where we saw problems in our (limited) testing was in the first 15nts off the sequencer.
