"averaging" replicates

I have a group that collected fecal samples with the omni-gut kit and gave me duplicates from each sample to process. I ran them through mothur treating each of the duplicates as its own sample. But they really want me to “average” the duplicates.

Would you do that in mothur or R? I can average the alpha diversity pretty easily in R, but am mentally stuck on how to average the lower triangle dist matrices. I’ve been thinking about converting from lt to 3 column dist, replace the sample name with replicate a or b with just sample name, then concat col1 and col2, average, split col1 and col2 apart again, then turn back into lt matrix. But that seems to be a not smart way to approach this.

If they’re technical replicates, then I would pool them either in make.contigs or after you’ve generated the shared file.


ok thanks. for future reference if someone has the same question-I’m going to merge.groups on the .shared file then rerun summary.single and dist.shared

as a philosophical question-do you view 2 independent extractions from a single fecal sample technical or biological replicates?

I’d treat them as technical replicates. Extraction kits being as expensive as they are, I’m not sure that I would ever do it though…

1 Like

I agree, this group thought their field wouldn’t accept the results without the replicates. But by averaging them, they’re not addressing variability within replicates so not sure how the extra expense is worth while. I’m running everything with the separate as well as averaged and I’ll see how they compare.

This topic was automatically closed 10 days after the last reply. New replies are no longer allowed.