[ae_ucla]

Yes, I have the same problem, I posted this in another thread:

http://seqanswers.com/forums/showthread.php?t=7109

[lshen]

I can get it working when using sorted SAM and sorted reference gene annotation to do quantification. But it seems declared too many nontest:

#Processed 22426 loci. [*************************] 100%
#Performed 4831 isoform-level transcription difference tests
#Performed 0 tss-level transcription difference tests
#Performed 3598 gene-level transcription difference tests

test_id gene locus sample_1 sample_2 status value_1 value_2 ln(fold_change) test_stat p_value significant
ENSCAFT00000000001 ENPP1 chr1:3251711-3321555 q1 q2 NOTEST 33.7819 1.42552 -3.16539 3.70202 0.000213891 no
ENSCAFT00000000003 - chr1:3363194-3365024 q1 q2 NOTEST 7.80975 3.21343 -0.888032 1.33992 0.180272 no
ENSCAFT00000000005 - chr1:3390440-3422494 q1 q2 NOTEST 0 0 0 0 1 no
ENSCAFT00000000006 PARD6G chr1:3508928-3565493 q1 q2 NOTEST 22.8188 0 6.95321e-310 2.22507e-308 0 no


It definitely should not be "NOTEST" with the FPKM values (26041 of 30913 isoforms are labeled as NOTEST").

But v0.9 did cut the de novo transcript prediction numbers a lot to about ~40000 from ~120,000. I am using single read data (84nt, 25M reads).

[Cole Trapnell]

I can get it working when using sorted SAM and sorted reference gene annotation to do quantification. But it seems declared too many nontest:

#Processed 22426 loci. [*************************] 100%
#Performed 4831 isoform-level transcription difference tests
#Performed 0 tss-level transcription difference tests
#Performed 3598 gene-level transcription difference tests

test_id gene locus sample_1 sample_2 status value_1 value_2 ln(fold_change) test_stat p_value significant
ENSCAFT00000000001 ENPP1 chr1:3251711-3321555 q1 q2 NOTEST 33.7819 1.42552 -3.16539 3.70202 0.000213891 no
ENSCAFT00000000003 - chr1:3363194-3365024 q1 q2 NOTEST 7.80975 3.21343 -0.888032 1.33992 0.180272 no
ENSCAFT00000000005 - chr1:3390440-3422494 q1 q2 NOTEST 0 0 0 0 1 no
ENSCAFT00000000006 PARD6G chr1:3508928-3565493 q1 q2 NOTEST 22.8188 0 6.95321e-310 2.22507e-308 0 no


It definitely should not be "NOTEST" with the FPKM values (26041 of 30913 isoforms are labeled as NOTEST").

But v0.9 did cut the de novo transcript prediction numbers a lot to about ~40000 from ~120,000. I am using single read data (84nt, 25M reads).

The NOTEST status is set when there are fewer than (by default) 500 reads in that locus in both samples. You should be able to resolve this by lowering this threshold with the -c option.

[lshen]

Thank you, Cole. I also figured out it be this reason after I sent my comment.

I like this version especially in that it now predicts much fewer transcripts with my single read data. Now the total number of predicted transcripts is in the range of known annotated transcripts. I will check the new predictions for validity. Thank you for the hard working to get the tool quickly improved.