JC, thanks a lot ! RPKM is exactly what I am trying to understand. :)
Let's say if I have to find "length per KB" of a read.
How should I go about it ?
I can find the "length" by subtracting the END coordinate with the read with the START coordinate.
Let's say Read 1 spans from 300 to 330. So, it's length is 30.
Does "Length per KB" mean I divide 30 by 1000 i.e 0.030 ?
2 answers
The "length per kilobase" as you pointed is the sequence size in kilobases units. But it's applied to gene sizes which spans for 1-2kb in average in the human genome, no reads that are really short. I think you are trying to understand RPKM see http://seqanswers.com/forums/showthread.php?t=586
the original concept applied to RNAseq is described in http://www.nature.com/nmeth/journal/v5/n7/full/nmeth.1226.html
Hello JC,
According to my understanding of the paper, the simplest way I can write a wordy formula for RPKM is :
RPKM = (sum of all reads mapping to all exons of the gene) / ((sum of lengths of all exons of the gene / 1000) * (sum of all reads mapping to all genes in the sample / 1000000))
Is that correct ?
However, if I have to find the RPKM for each exon of each gene, would I modify the formula to :
RPKM = (total #reads mapping to the exon) / ((length of exon / 1000) * (sum of all reads mapping to all genes in the sample / 1000000))
Please correct if I am going wrong.
Thanks.
You are right in both formulas. Recently, bioconductor has a new package to deal with RNAseq data in exons: DEXSeq http://www.bioconductor.org/packages/devel/bioc/html/DEXSeq.html Paper: http://precedings.nature.com/documents/6837/version/2/files/npre20126837-2.pdf
Log in to answer this question.