Dear Delong, Thanks for the reply. Your explanation totally makes sense, but I'm asking about something else. If you look at the annotation files, the two isoforms do not overlap at all (they are alternative last exons) in the examples, so it is not the matter of deciding one isoform versus the other. Other than that the assignment of reads to isoforms are more or less similar (the ratio), so I don't think your example could describe why the differences in PSI values happen.
Best, Alborz On Thu, Oct 1, 2015 at 9:20 AM, Zhou Delong <[email protected]> wrote: > Hello, > > Yarden or anybody on the ML, please correct me if I'm wrong but here is my > explaintion for part of your question: > > Say we have a pair of reads that each of them is within one exon: left in > e1, right in e2 (cf attached) > > If you use single end mode, the left read will support two isoforms > (e1-e2-e3 & e1-e3) while the right read will support the long isoform only. > > However in paired end mode, there is only one read that supports only the > long isoform. > > In fact, when you use single end mode for paired end sequencing, you lose > the information that the two reads are linked and come from the same > transcript. This can can affect reads assignment like in the case I > described above and other sorts of cases. > > As why most affected events have a huge difference in isoform length, I > guess longer each exon, larger the chances you got something like my > exemple happenning in your reads. > > Best, > > Delong > > > ------------------------------ > *De :* [email protected] <[email protected]> de la part > de Alborz Mazloomian <[email protected]> > *Envoyé :* 30 septembre 2015 18:59 > *À :* [email protected] > *Objet :* [miso-users] paired end vs. single end modes > > Dear Yarden, > > I have a question regarding paired end versus single end options. I have > paired end libraries and I use the annotation files that are provided in > Miso webpage. > > As suggested in the documentation (possibility of missing events having > short length in paired end mode), I tried both single end and paired end > modes. I realized that single end mode could detect more events with a > descent overlap with paired end results (considering diff cutoff of 0.1 for > both options) > > However, when I check the events that are detected in paired-end mode but > not in single end mode, I find that they have big bayes factors, but the > difference in PSIs is small. It mostly happens for regions where there is a > huge difference between isoform lengths (isoform1 vs. isoform2). Please > have a look at the two examples below: > -------------------------------------------------- > Example 1: > annotation: > chrX ALE gene 54094757 54200579 . - . > [email protected]@uc004dsz.2;[email protected] > @uc004dsz.2;[email protected]@uc004dsz.2 > chrX ALE mRNA 54094757 54099720 . - . > [email protected]@uc004dsz.2.B;[email protected] > @uc004dsz.2;[email protected]@uc004dsz.2.B;[email protected] > @uc004dsz.2 > chrX ALE exon 54094757 54099720 . - . > [email protected]@uc004dsz.2.B.0;[email protected] > @uc004dsz.2.B;[email protected]@uc004dsz.2.B.0;[email protected] > @uc004dsz.2 > chrX ALE mRNA 54200222 54200579 . - . > [email protected]@uc004dsz.2.A;[email protected] > @uc004dsz.2;[email protected]@uc004dsz.2.A;[email protected] > @uc004dsz.2 > chrX ALE exon 54200222 54200579 . - . > [email protected]@uc004dsz.2.A.0;[email protected] > @uc004dsz.2.A;[email protected]@uc004dsz.2.A.0;[email protected] > @uc004dsz.2 > > paired end: > [email protected]@uc004dsz.2 0.98 0.93 1.0 0.42 0.35 > 0.5 0.56 1000000000000.0 ('[email protected]@uc004dsz.2.B.0', > '[email protected]@uc004dsz.2.A.0') (0,0):1063,(1,0):1494 0:1494 > (0,0):603,(0,1):41,(1,0):855 0:855,1:41 chrX NA (54094757, > 54200222) (54099720, 54200579) > > single end: > [email protected]@uc004dsz.2 1.00 1.00 1.00 0.94 0.93 > 0.95 0.06 1000000000000.00 '[email protected] > @uc004dsz.2.B.0','[email protected]@uc004dsz.2.A.0' > (0,0):2168,(0,1):6,(1,0):30720:3072,1:6 > (0,0):1245,(0,1):109,(1,0):1746 0:1746,1:109 chrX NA > 54094757,54200222 54099720,54200579 > > ---------------------------------------------- > Example 2: > chr9 ALE gene 126141933 126164191 . - . > [email protected] > @uc010mwh.1uc004bny.1uc004bnz.1;[email protected] > @uc010mwh.1uc004bny.1uc004bnz.1;[email protected] > @uc010mwh.1uc004bny.1uc004bnz.1 > chr9 ALE mRNA 126141933 126144904 . - . > [email protected] > @uc010mwh.1uc004bny.1uc004bnz.1.B;[email protected] > @uc010mwh.1uc004bny.1uc004bnz.1;[email protected] > @uc010mwh.1uc004bny.1uc004bnz.1.B;[email protected] > @uc010mwh.1uc004bny.1uc004bnz.1 > chr9 ALE exon 126141933 126144904 . - . > [email protected] > @uc010mwh.1uc004bny.1uc004bnz.1.B.0;[email protected] > @uc010mwh.1uc004bny.1uc004bnz.1.B;[email protected] > @uc010mwh.1uc004bny.1uc004bnz.1.B.0;[email protected] > @uc010mwh.1uc004bny.1uc004bnz.1 > chr9 ALE mRNA 126163870 126164191 . - . > [email protected] > @uc010mwh.1uc004bny.1uc004bnz.1.A;[email protected] > @uc010mwh.1uc004bny.1uc004bnz.1;[email protected] > @uc010mwh.1uc004bny.1uc004bnz.1.A;[email protected] > @uc010mwh.1uc004bny.1uc004bnz.1 > chr9 ALE exon 126163870 126164191 . - . > [email protected] > @uc010mwh.1uc004bny.1uc004bnz.1.A.0;[email protected] > @uc010mwh.1uc004bny.1uc004bnz.1.A;[email protected] > @uc010mwh.1uc004bny.1uc004bnz.1.A.0;[email protected] > @uc010mwh.1uc004bny.1uc004bnz.1 > > > paired end: > [email protected]@uc010mwh.1uc004bny.1uc004bnz.1 0.36 > 0.3 0.44 0.68 0.570.78 -0.32 12557315.78 > ('[email protected]@uc010mwh.1uc004bny.1uc004bnz.1.B.0', > '[email protected]@uc010mwh.1uc004bny.1uc004bnz.1.A.0') > (0,0):552,(0,1):49,(1,0):587 0:587,1:49 > (0,0):275,(0,1):17,(1,0):770 0:770,1:17 chr9 NA (126141933, > 126163870) (126144904, 126164191) > > single end: > [email protected]@uc010mwh.1uc004bny.1uc004bnz.1 0.89 > 0.88 0.91 0.97 0.960.98 -0.08 1000000000000.00 > '[email protected] > @uc010mwh.1uc004bny.1uc004bnz.1.B.0','[email protected]@uc010mwh.1uc004bny.1uc004bnz.1.A.0' > (0,0):1119,(0,1):148,(1,0):1228 > 0:1228,1:148(0,0):523,(0,1):50,(1,0):1605 0:1605,1:50 chr9 NA > 126141933,126163870 126144904,126164191 > --------------------------------------------------- > > I am using Miso version 0.4.9. Could you please help me figure out why > this happens in single end mode? > > Thanks, > Alborz > > >
_______________________________________________ miso-users mailing list [email protected] http://mailman.mit.edu/mailman/listinfo/miso-users
