Unique mapper on genome - multi mapper on transcriptome

[StephaniePi83]

Hi everybody,
I'm analyzing iCLIP data (technique to study protein-RNA interactions on a genome-wide scale). I have two populations bound by the protein of interest - small RNA and mRNA.
I take all reads that map uniquely to the genome and i wanted to know all transcript that are bound by our protein of interest.
But i have a problem, i use bowtie to map unique mapper reads on the genome tothe transcriptome, but unique mapper on the genome became multi mapper on transcriptome, i'm confused, how to deal with it ? I have some very small reads ( the smallest are 13 nt).
How can i assign reads to transcript without mistake, i mean how to deal with reads aligning to multiple genes ?
do i have to associate each read with all transcripts it aligns to?
Thanks in advance for your help.

[paulsydney]

Hi Stephanie,

I wonder if you have published your iCLIP study recently. I am keen to learn about how to analyse iCLIP data.

[StephaniePi83]

Hi paulsydney,
we haven't published yet.
Some paper help me :

http://www.ncbi.nlm.nih.gov.gate1.inist.fr/pubmed/21048981

http://www.ncbi.nlm.nih.gov.gate1.inist.fr/pubmed/20601959

http://www.ncbi.nlm.nih.gov.gate1.inist.fr/pubmed/21358640

[paulsydney]

Thanks Stephanie. I am still trying to decode many parts of the analysis section of that paper. Did you end up using any other bioinformatics tools apart from bowtie? I am not to certain how to extract annotation data from my map.

[StephaniePi83]

After removing adapter from my reads, i mapped all read on the genome allowing up to 2 mismatches with bowtie. I translate bowtie output file to bed file; then i use Bedtools (intersectBed function) to annotate all reads.

[paulsydney]

Hi Stephanie,

What tools did you use to randomise iCLIP position when doing the pentamer z-score analysis?

[StephaniePi83]

Hi Paul,
I didn't do this analysis neither the statistical analysis. My supervisor send our data to a person who did a pipeline to analyse iclip data (he did all bioinformatic analysis in the paper i send you).

[kopi-o]

Unless I'm misunderstanding something, there isn't anything weird about unique genome mappers that are also multiple transcriptome mappers. Consider a read that spans a splice junction and can be uniquely mapped to the genome. If the gene containing the splice junction has multiple isoforms containing that splice junction, the read will map to multiple isoforms in the transcriptome reference. Does that answer the question?

[kopi-o]

Unless I'm misunderstanding something, there isn't anything weird about unique genome mappers that are also multiple transcriptome mappers. Consider a read that spans a splice junction and can be uniquely mapped to the genome. If the gene containing the splice junction has multiple isoforms containing that splice junction, the read will map to multiple isoforms in the transcriptome reference. Does that answer the question?

[paulsydney]

Ah ic... That was a great idea. I should suggest that to my supervisor too actually.

[StephaniePi83]

Hi kopi-o,
This is what i was thinking ! it map to all transcript of the same gene on transcriptome, and some reads will map to exon-exon juction that are not present on the genome.