Chimera checking after using swarm to pick otus

[Jessica Hardwicke]

I'm trying to understand how to utilize swarm and vsearch with my qiime pipeline (paired-end illumina seqs) and I'm a little confused. Here's what I think this part of my pipeline should be:

1. split libraries (qiime)

2. dereplecation (vsearch) 

3. pick otus (qiime, using swarm)

4. remove chimeras (vsearch)

I'm just unsure how to go from #3 to #4 - As far as I know vsearch input should be a fasta file, but that's not the output of pick otus (it's a txt file representing clusters). I've seen some discussions online but not any so far that use qiime/swarm and I'd like to keep things mostly in qiime if I can. 

[zhiying Guo]

Hello,

Why do you think chimera-filter should be after picking otus? In reality, you can change the order of step3 and step4. 

As I know, Uparse pipeline and RDP pipeline usually remove chimeras before clustering (pick_otus).

Best,

Zhiying

[Jessica Hardwicke]

I got this information from the swarm wiki:

When clustering with swarm, when is an appropriate time to check for chimeras?

After clustering.

As pointed out by Colin Brislawn, Robert Edgar recommends using uchime to remove chimeras after OTU picking with uparse. His recommendation is based, in part, on chimera removal heuristics built in to uparse. When using uclust, Robert Edgar recommended referenced-based chimera checking before OTU picking so that OTUs were not influenced by chimeras. When using swarm with default parameters or low d values (i.e. high-resolution clustering), chimeras and non-chimeras should form independent OTUs. It is therefore more interesting to perform the chimera detection on OTU representatives, as the dataset size reduction allows to perform both de novo-and reference-based approaches.

[zhiying Guo]

Hello,

Thanks for your information. Sorry that I don't much knowledge about swarm. According to wiki, you can run the pipeline as step1~ step4.

Because chimera and non-chimeras will cluster independent OTU, that means you can pick representative sequence (rep_set) after OTU clustering. 

According to the result of pick_otus.py, use pick_rep_set.py to pick rep_set, use this rep_set (fasta file) as input file of vsearch to filter chimeras.

I have run swarm on my data with QIIME 1.91. successfully got rep_set of each OTU.

command is below,

pick_otus.py -i seqs.fna -o otus_swarm -m swarm

pick_rep_set.py -i ./otus_swarm/seqs_otus.txt -f seqs.fna -o rep_set.fna

But I don't have vsearch. so I cant't run the total pipeline.

Best,

Zhiying

[AS]

Can someone please tell me if de-novo chimera checking using vsearch can be conducted on representative sequences generated after picking OTUs with sumaclust?

[TonyWalters]

Hello,

Standard disclaimer: QIIME1 is no longer being updated, and it's recommended to move to QIIME2:

https://qiime2.org/
(if you did this, you would want to start with your raw reads for quality/chimera checking)

For your question in qiime1, unfortunately, you would need the pre-clustered sequences to use vsearch.

For post-OTU picking sequences, you could follow the ChimeraSlayer approach:

http://qiime.org/tutorials/chimera_checking.html

-Tony