chimera.uchime abskew

Does the abskew parameter make sense when it comes to low abundance sequences?

When using uchime with the self reference option it relies on the abskew parameter.

" The abskew parameter can only be used with template=self. Minimum abundance skew. Default 1.9. Abundance skew is: min [ abund(parent1), abund(parent2) ] / abund(query). "

This is to make use of the idea that chimeras are always less common than their parents, but I figure there are limitations to the way this parameter is formulated when it comes to low abundance parents and queries.

Actually I observe a lot of cases in my data like this: abundance parent 1= 20, abundance parent 2 = 2, abundance query =1. This fulfills the abskew criterion of 1.9 since parent 2 is twice as common as the query. In other words many singleton sequences are classified as chimeras of very low abundance parents. How likely is it that there is a chimeras derived from a parent at whose frequency is 2 in 5000 sequences ?
I am wondering if I am largely underestimating diversity in my sample because of the chimera search. My feeling is that I am rather dealing with a lot of closely related sequences in my dataset that are not necessarily chimeras. I used precluster before runnning uchime.

Does anyone have an opinion or suggestions?

Thanks,
Fabian

Yeah, I wondered about that too. Because PCR isn’t 100% efficient, I think the 2-fold difference might be too conservative. I haven’t pushed it because it is really hard to test this (difficult to come up with good testing sets). I’d encourage you to check with Robert Edgar to see the extent of his testing on this.

Since my question is chimera.uchime abskew related I thought would be better to ask it here.

It seems at this step mothur only uses 1 cluster; It says if abwkew=self, it only uses 1 cluster; and in the mothur document its stated the abskew parameter should set to self. I also played with the abswkew parameter, but seems the value does not have any effect on the number of clusters.

so I appreciate if you make a comment how can I facilitate this step.

I think you mean reference= instead of abskew= right?

it says
Checking sequences from /eva/users/djf482/Projects/MS96/output/stability.contigs.good.unique.good.trim.good.filter.unique.precluster.fasta …
When using template=self, mothur can only use 1 processor, continuing.
uchime v4.2.40

and I can not change the template ! the uchime inside mothur only uses 1 cluster and is very slow compare to the original uchime ! I appreciate any comment on how to boost up its speed.

Thank you

From the MiSeq SOP, http://www.mothur.org/wiki/MiSeq_SOP

chimera.uchime(fasta=stability.trim.contigs.good.unique.good.filter.unique.precluster.fasta, count=stability.trim.contigs.good.unique.good.filter.unique.precluster.count_table, dereplicate=t, processors=2)

Mothur will process each group individually and use multiple processors.

what it means then

uchime by Robert C. Edgar
http://drive5.com/uchime
This code is donated to the public domain.

Checking sequences from /eva/users/djf482/Projects/MS96/output/stability.contigs.good.unique.good.trim.good.filter.unique.precluster.fasta …
When using template=self, mothur can only use 1 processor, continuing.
uchime v4.2.40
by Robert C. Edgar
http://drive5.com/uchime
This code is donated to the public domain.

It seems to me it uses only 1 processor eventhough I have set the number of processors to 64.

What is the exact command syntax that you are using?

Hi,

I think I know why chimera.uchime uses only one processor.

It also happened to me, and later I realized that when I created the count table, using count.seqs, I forgot to add the group file.

When I redid the command with a count_table containing the group counts, the chimera.uchime used more than one processors.

Cheers,

Maya