Is there anyone knows how to calculate FPKM for specific UTR region?

[zjrouc]

Is there anyone knows how to calculate FPKM for specific UTR region or just specif exon?

[dpryan]

The same as you would for anything else. You take the number of fragments mapping to each region, divide that by the length of each region in kilobases, and then divide the results by a million.

[zjrouc]

Then how do you count reads in that region? Manually?

[dpryan]

You have a plethora of choices: htseq-count, featurecounts, summarizeOverlaps (in R), bedtools intersect, samtools view (note the -c option), etc.

[zjrouc]

So if i want to use htseq to count reads for this specif region. i need to generate a gtf file for this region, right? now i have problem with that. my htseq or cufflink results from using this own gtf file were empty. i am wondering if there are other easier ways

[dpryan]

Yes, htseq-count requires a GTF file. If you can see alignments in IGV and you don't get any with the GTF file and htseq-count then you're probably doing something wrong. Try making a BED file and use bedtools intersect with that and the GTF file separately. If they both work and give approximately the same results then you know that you were using htseq-count wrong. If they give vastly different results then there's likely something wrong with the GTF file. If they don't give any alignments then you've screwed up specifying the region.