Hi Yarden,

Thanks for the explaining below. Since the assigned count is not an 
interpretable quantity for any downstream analyses, is it reasonable to use the 
read class counts assigned to (1,0) and (0,1)? Does the filtering script 
filter_events actually using the read class counts instead of assigned counts?

Thank you,
Xiaojia Tang


-----Original Message-----
From: [email protected] [mailto:[email protected]] On Behalf 
Of Yarden Katz
Sent: Thursday, July 07, 2016 4:33 PM
To: Wang, Ziheng
Cc: [email protected]
Subject: Re: [miso-users] Question about definition of assigned counts

The assigned counts parameter should not be used for anything; it's there only 
for internal debugging purposes.  It is the assignment of reads at the last 
iteration of the Monte Carlo simulation which is not an interpretable quantity 
for any downstream analyses.  If you want to get an expected number of reads 
assigned to each isoform, you can use the Psi values and the length of the 
isoforms for this.

To answer your question about RPKM: if you're using an exon-centric annotation 
(e.g. skipped exon trios), it would not be a good idea to use this to calculate 
RPKMs.  The reason is that there's no guarantee that the flanking exons are 
constitutive relative to the whole gene.  You can have an alternative exon with 
an upstream constitutive exon, and a downstream alternative exon (the latter 
could be "the skipped exon" in another event in the annotation.)  Even if the 
two flanking exons of the alternative exon of interest were constitutive, this 
would be a poor estimate of RPKM since it only relies on two exons (using other 
constitutive internal exons and long constitutive regions of the 3' UTR would 
give a much better estimate.)  In short, estimating RPKM based on constitutive 
exons is much easier than estimating isoform abundance, so it's best not to 
convert an easier problem into a much harder one. 

Yarden

> On Jul 7, 2016, at 5:03 PM, Wang, Ziheng <[email protected]> wrote:
> 
> Dear MISO users,
> 
> I am just wondering what the assigned counts parameter in the summary file is 
> consisted of. Are constitutive reads from the exon regions included in this 
> number? If I sum all the assigned counts up, can I theoretically calculate an 
> RPKM for the gene?
> 
> Thank you,
> 
> Ziheng Wang
> _______________________________________________
> miso-users mailing list
> [email protected]
> http://mailman.mit.edu/mailman/listinfo/miso-users


_______________________________________________
miso-users mailing list
[email protected]
http://mailman.mit.edu/mailman/listinfo/miso-users


_______________________________________________
miso-users mailing list
[email protected]
http://mailman.mit.edu/mailman/listinfo/miso-users

Reply via email to