Mapping of Paired end sequenced Amlicon?

[jeni]

Hi guys,

I am puzzled how to do mapping of my amplicons PE seq reads. Library was prepared in such a way that PE seq has been done from both sides of amplicons. Now I am slightly confused, How to map these reads over the reference amplicons using bowtie?

[Brian Bushnell]

Is there a particular reason you need to use bowtie?

[jeni]

No there is no perticular reason. I could use other tool as well.

My question is related with the mapping. If it is single end, I can map it easily over the reference amplicon but how to map the Paired reads over the amplicons?

[GenoMax]

NGS aligners are able to account for the paired-end reads. You would have no problems as long as you properly indicate input files with read pairs being mapped. What size amplicons do you have?

[jeni]

I have reads which are around 200 bp long.

[GenoMax]

See BBMap and the example for how to use paired end reads for mapping.

[jeni]

Merge overlapping PE reads

In continuation of the question, I have another idea of merging the overlapping reads of the PE sequenced fastq data.

I found a very interesting page related to this: http://thegenomefactory.blogspot.com...aired-end.html

They have suggested few tools. eg. fastq-join, FLASH etc. Have anybody tried any of such tools? So that I can have a review.


I would appreciate if any one could suggest me for the PE sequenced data,
1) Should I directly go for Mapping using PE option (eg. bowtie1) OR
2) merge the PE reads R1 and R2 together to make a super-long SE read, with extra confidence of the middle bases from consensus of the overlapping sections of R1 and R2?


Thanks all in advance.

[GenoMax]

If you know that your reads are overlapping then you can merge them into a longer representation before mapping. If you find that the reads are not merging well (because insert size was bigger than you expected) then you should proceed with mapping them as paired-end.

BBMap suite has a tool that does the merging called BBMerge. FLASH is another well regarded tool for merging reads.

[jeni]

Thanks for your response.

I have ~350bp long DNAs fragments and after PE sequencing the length of reads are 200 (FastQC). I would expect an overlap of ~50 bases. And that is the reason, I am thinking of merging not directly mapping with PE parameter.

[Brian Bushnell]

Both approaches are valid, just bear in mind that not all PE reads will merge even if they overlap, particularly if are low-complexity in the overlapping region or if the reads have high error rates in their tails. Merged reads are typically easier to work with, but unmerged reads may have lower bias if a large percentage do not merge successfully.

It's also possible to map paired reads, merge them based on mapping coordinates, then remap the merged reads.

Is your goal variant calling, quantification, or something else?

[jeni]

Basically quantification.

[Brian Bushnell]

For quantification, I recommend mapping the unmerged reads as pairs, to prevent bias against low-complexity sequence.