Hi All,

I encountered several error messages during cuffdiff run. The error message is as follows:

[13:43:11] Loading reference and sequence.

No fasta index found for ../ref/all.fa. Rebuilding, please wait..

Fasta index rebuilt.

[13:44:04] Inspecting maps and determining fragment length distributions.

[13:46:04] Calculating initial abundance estimates.

> Processed 23484 loci. [*************************] 100%

[13:51:17] Learning bias parameters.

[13:53:15] Testing for differential expression and regulation in locus.

> Processing Locus chr1:5441872-5445261 [ ] 2%Error: sqrt(det(cov)) == 0, 0.000000 after rounding.

Error: sqrt(det(cov)) == 0, 0.000000 after rounding.

> Processing Locus chr10:18647262-18648745 [**** ] 17%Error: sqrt(det(cov)) == 0, 0.000000 after rounding.

Error: sqrt(det(cov)) == 0, 0.000000 after rounding.

> Processing Locus chr10:19656777-19663522 [**** ] 17%Error: sqrt(det(cov)) == 0, 0.000000 after rounding.

Error: sqrt(det(cov)) == 0, 0.000000 after rounding.

> Processing Locus chr10:20340517-20356103 [**** ] 18%Error: sqrt(det(cov)) == 0, 0.000000 after rounding.

Error: sqrt(det(cov)) == 0, 0.000000 after rounding.

This error did not halt the program and cuffdiff actually kept running until all results were output. I am not sure what did those errors mean? Are they fatal errors that might lead to the error in my result.

Here is what I did before running cuffdiff.

I ranthe Tophat to get my alignment results "sample_X/accepted_hits.bam" (I have 10 samples here, so X = 1..10). I ran Cufflinks to get the "sample_X/transcripts.gtf". Then I use Cuffcompare using all transcripts.gtf files output by cufflinks to get a combined gtf file. At the end, I feed this combined gtf file into cuffdiff and two samples from my sample pool to do the diff-express analysis.

Thanks,

Zheng

I encountered several error messages during cuffdiff run. The error message is as follows:

[13:43:11] Loading reference and sequence.

No fasta index found for ../ref/all.fa. Rebuilding, please wait..

Fasta index rebuilt.

[13:44:04] Inspecting maps and determining fragment length distributions.

[13:46:04] Calculating initial abundance estimates.

> Processed 23484 loci. [*************************] 100%

[13:51:17] Learning bias parameters.

[13:53:15] Testing for differential expression and regulation in locus.

> Processing Locus chr1:5441872-5445261 [ ] 2%Error: sqrt(det(cov)) == 0, 0.000000 after rounding.

Error: sqrt(det(cov)) == 0, 0.000000 after rounding.

> Processing Locus chr10:18647262-18648745 [**** ] 17%Error: sqrt(det(cov)) == 0, 0.000000 after rounding.

Error: sqrt(det(cov)) == 0, 0.000000 after rounding.

> Processing Locus chr10:19656777-19663522 [**** ] 17%Error: sqrt(det(cov)) == 0, 0.000000 after rounding.

Error: sqrt(det(cov)) == 0, 0.000000 after rounding.

> Processing Locus chr10:20340517-20356103 [**** ] 18%Error: sqrt(det(cov)) == 0, 0.000000 after rounding.

Error: sqrt(det(cov)) == 0, 0.000000 after rounding.

This error did not halt the program and cuffdiff actually kept running until all results were output. I am not sure what did those errors mean? Are they fatal errors that might lead to the error in my result.

Here is what I did before running cuffdiff.

I ranthe Tophat to get my alignment results "sample_X/accepted_hits.bam" (I have 10 samples here, so X = 1..10). I ran Cufflinks to get the "sample_X/transcripts.gtf". Then I use Cuffcompare using all transcripts.gtf files output by cufflinks to get a combined gtf file. At the end, I feed this combined gtf file into cuffdiff and two samples from my sample pool to do the diff-express analysis.

Thanks,

Zheng

## Comment