3'UTR length from RNA-seq

[Palgrave]

Hi all,

Is there a way the measure 3'UTR length from RNA-Seq Data?
And does there exist GTF tracks containing genes with sorted 3'UTR length?

Best

[sahilsukla]

Shorted 3' UTR length analysis in RNA-Seq data

Hi Palgrave,

Did you get any answer for this query????

Actually I am looking for some tool or pipeline to identify shorted 3' UTR from RNA-Seq data. I have illumina HiSeq 2000 100 base PE data for human.

Best

[Palgrave]

No I did not. Maybe anyone knows??

[syfo]

Unless the protocol was specifically designed to identify mRNAs 3' ends, each RNA-seq read can originate from anywhere within a (processed) transcript. Identifying 3'UTRs is not a trivial task. You can look for the reads that map into known 3'UTRs by comparing the alignment file with a gene annotation though (BEDtools, HTSeq etc).
You can download a GFF or BED file with annotated 3'UTRs from the UCSC table browser for instance, given that your species of interest is available. Select the genes and gene predictions category, BED format, then on the second page you can specify "3'UTRs".

[bruce01]

Well, UTR from RNAseq data seems like an oxymoron to me?

If you have 100bp PE and it overlaps to 3' UTR, you could try making your own UTR GTF from Biomart? Then use this to guide a UTR-specific alignment?

[syfo]

Well, UTR from RNAseq data seems like an oxymoron to me?

If you have 100bp PE and it overlaps to 3' UTR, you could try making your own UTR GTF from Biomart? Then use this to guide a UTR-specific alignment?

Delineating the "3'UTRome" with RNA-seq is not impossible: see here and here for instance. These were specific protocols though.
From a standard RNA-seq data you can still try to get an idea about potential 3'UTRs extension by looking at the reads that overlap annotated 3'UTRs and/or map downstream of the polyadenylation point as you suggest, however keep in mind that each read can come from a different transcript -and even from a different gene- in which this position might be coding. In other words, a same genomic region can be annotated as UTR, CDS, intronic etc depending on the annotated mRNA you consider -not mentioning overlapping genes.
Anyway, coming back to the original question ("can you estimate 3'UTR length from RNA-seq data") my answer would be "no".

[bruce01]

So it's called "3P-seq", a new one on me! If you were strict enough with the alignment from a custom-made GTF that only took a certain amount of 3' it might work, but these sorts of things (in my experience) give very uncertain answers.

[Palgrave]

My hypothesis is that longer 3'UTR-isoforms are down-regulated in treated cells to escape miRNA targeting.

I have require the FASTA sequence and the Coordinates for all RefSeq 3'UTR from UCSC. Perhaps one could see if there is a significant difference between the number of reads that map to start of the 3'UTR compared to the end?


Maybe cufflinks could be used to look for isoforms, and then group these isoforms with regard to the 3'UTR?

[syfo]

I would start by considering the already annotated isoforms that differ by their 3'UTRs (if any).