problem with 16S reverse reads with low quality

Question

[michele.lamprillo]

Hi everybody,

i'm doing an analysis on 16S PE amplicon sequencing data of soil samples taken from the same soil in two different seasons. I want to measure and test for differences in alpha diversity using 3 samples per season.

My problem is that i have one of these samples (i'm gonna call it A) already with lower number of reads, than also the reverse (R2) reads of this sample are low quality!! ... so when i use trimmomatic the number of total reads of A is already reduced (around 48% of only R1 surviving reads). after all of this i have around 17k of sequences for R1A and R2A while around 100k for R1 and R2 in the other samples.

Wouldn't it be better if i just remove R2A prior to trimmomatic so i'll reatin more information?

do you think reviewer will be ok with this? (treating different samples in different ways)

Michele

Answer 1

[hakimbazol]

First of all, if you still have the samples and the funding, I recommend you to re-sequencing them again, if you have enough time.

However, in this case, I assume your sample is 16s, then it should be bacteria. You may use only R1 as long you have the phenotype result, for example increment of pH due to nitrite amount to prove your soil sample contains nitrite-synthesis bacteria. Then, all reads should be treated similar as your worst case.

Hope it helps!

[michele.lamprillo]

Thank you,

Unfortunately i can't re-sequence any of my samples.

---

However, in this case, I assume your sample is 16s, then it should be bacteria. You may use only R1 as long you have the phenotype result, for example increment of pH due to nitrite amount to prove your soil sample contains nitrite-synthesis bacteria. Then, all reads should be treated similar as your worst case.

Ok, so, if i'm not wrong, you mean i should only retain R1 also for the other samples where R2 is not bad? ..but this way i'm gonna lose information of paired reads in the good samples!

[hakimbazol]

Yes, I understand this issue. However, another option is to remove the problematic sample anyway, and discussing the result with the left samples. This is other way that can still maintain the quality unless there is a bad sample in your case.