mapping on duplication gene?

[g781]

Hi anyone,

Are any suggestions for mapping on duplication gene using short reads (50bp paired-end)?

I have several RNAseq data (paired-end with 50bp) for Tbrucei. Tbrucei is exon-less. Almost all of genes have only one exon. For gene with 2 more exons, the distance between exons is very short, being less than 10bp in average. So I don't have to consider the splice problem.

My problem is that Tbrucei has a lots gene duplication (roughly ~200). It's very hard to count how many short-reads are on gene duplication, so I've ignored by far those gene duplication.

Has anyone had the same problem with me? or is it still a very hard challenge except for using long-reads (e.g. 454).

Any suggestions will be appreciated.

[philippd]

Hi g781,

I think it depends on how similar your duplicated genes are and what kind of analysis you would like to perform. In general after mapping the reads to duplicated genes you have two types of reads. Those that map perfectly to all duplications and those that map better to one of the duplication than to the others.

The fraction of the two types of reads depends on how similar the duplications are.
If they are identical then it will be very hard (if not impossible) to resolve the reads that map to multiple loci. If they are more different then it becomes easier.

One strategy I used once to detect differentially expressed genes when having duplicated genes, was to ignore all reads which map to two different locations with a similar score. That is you take a read only for you analysis if it maps to only one location very good and it the next best mapping is significantly worse.
This is of course problematic when you remove all the reads mapping to a gene and you would have to decide what to do in those cases.

[g781]

Hi Philippd,

Thanks for your suggestion.

Unfortunately, I was briefly to check about some of duplicated genes, its sequences are 100% the same. Basically, I don't have no a list of duplicated genes. what i known duplicated genes is that I looked back those duplicated genes from plotting the RANseq reads filtered by samtools with -f2 -q20. I found some of genes are no signals or weaker signals. after check about those genes' sequencing, its are all almost 100% the same. Few genes are partially different, so I still could detect the signals of those genes as you mentioned.

Today, I told with one guy who used to map reads on duplicated genes randomly. This way, in theory, is to divide signal by number of gene duplications . I just thought this problem seems to able be resolved only by using long-read so that it can cover to different region without repeat.

Thanks Philippd again.