saharbagheri
saharbagheri
I see. Thanks!!! So if we keep the randomize=T and by any chance the sample or samples from which reads were used to learn errors were under representative and results...
Hi, I discovered that the discrepancy mentioned above had arisen because I had used different subsampling percentages across projects, causing the “error-learnt” object to differ in the above example. Interestingly,...
Thanks for clarifying! It sounds like using most (or all) reads from a smaller number of samples can be more effective than simply increasing the total number of samples. However,...
Great! Thank you so much for getting back to me so quickly. I really appreciate it.
My final question is whether there is an effective solution for learning the error model and subsequently inferring microbial composition when we have samples from different body sites of the...
Thanks for your prompt reply and suggested solution. This is paired-end data. The spacers (with different length) are before the primers (in a way being partial), do they still get...
Hi again, Thanks for your explanation. I ended up updating my cutadapt to v5.1 and then: fwd_primer: "ACGTAC**ACTCCTRCGGGAGGCAGCAG**" (longest sequence, containing the longest spacer+primer) rev_primer: "ATCATG**GGACTACHVGGGTWTCTAAT**" (longest sequence, containing the...
Hi! To follow up on this I also figured out that some ASVs at the end still had the primers in them but they were usually low abundant, annotated as...
Thanks for your detailed responses! I apologize for not pasting the reads in a more readable format.