Tweet
Problem: cufflinks hangs/freezes during processing with the terminal showing the following messages:
Inspecting reads and determining fragment length distribution.
> Processing Locus chr12:69154464-69154707 [*** ] 13%
The data is paired end RNA seq data from mouse. I used the mm10 index files available on the cufflinks website and the following tophat and cufflinks commands:
> tophat
>cufflinks
I have read various related posts on this site including the following:
Using the suggestions from these posts, I created a 'Masks.gtf' file which includes all the rRNA, tRNA and mitochondrial regions. I then added the following to the cufflinks parameters:
Unfortunately the problem remained. I double checked the format of the masks file, used verbose mode make sure there were no hidden error messages, etc.
Next, I uploaded read coverage data to the UCSC browser and examined the read coverage near the offending region (chr12:69154464-69154707) and found that there two were large spikes in coverage in two subsequent regions on chr12, roughly at positions 69158856...69160012 and 69360961...69361955. So, I added the following three lines to the top of my Masks.gtf file, to see if I could force these regions to be excluded.
chr12 BEH exon 69155265 69159154 . + . gene_id "beh0"; transcript_id "beh0_region";
chr12 BEH exon 69158856 69160012 . + . gene_id "beh1"; transcript_id "beh1_region";
chr12 BEH exon 69360961 69361955 . + . gene_id "beh2"; transcript_id "beh2_region";
This did not have any noticeable effect; the program hangs in the exact same position, with the same messages.
The last thing I tried, was to try to make the program skip all 'bundles' with large numbers of aligned reads, by using the "--max-bundle-frags" flag. I tried the following scenarios, all with no noticeable effect:
In fact, when I examined the output using the -v flag, cufflinks gave no indication that it was skipping bundles with out-of-range read counts. E.g. the output contained messages such as the following even when --max-bundle-frags was equal to 100:
All of the above were tried on two different versions of cufflinks (v2.0.2 and v1.3.0) on a 64 bit linux computer.
Does anyone have any other advice for me? It seems like cufflinks simply won't work on my data set.
Inspecting reads and determining fragment length distribution.
> Processing Locus chr12:69154464-69154707 [*** ] 13%
The data is paired end RNA seq data from mouse. I used the mm10 index files available on the cufflinks website and the following tophat and cufflinks commands:
> tophat
-p 8
-G genes.gtf
-o tophat_output_directory
/mm10/Sequence/BowtieIndex/genome
/Working/WT_PE1.fastq
/Working/WT_PE2.fastq
-G genes.gtf
-o tophat_output_directory
/mm10/Sequence/BowtieIndex/genome
/Working/WT_PE1.fastq
/Working/WT_PE2.fastq
>cufflinks
-p 8
-o cufflinks_output_directory
tophat_output_directory/accepted_hits.bam
-o cufflinks_output_directory
tophat_output_directory/accepted_hits.bam
I have read various related posts on this site including the following:
http://seqanswers.com/forums/showthread.php?t=12458
http://seqanswers.com/forums/showthread.php?t=7169
http://seqanswers.com/forums/showthread.php?t=7169
Using the suggestions from these posts, I created a 'Masks.gtf' file which includes all the rRNA, tRNA and mitochondrial regions. I then added the following to the cufflinks parameters:
-M Masks.gtf
Unfortunately the problem remained. I double checked the format of the masks file, used verbose mode make sure there were no hidden error messages, etc.
Next, I uploaded read coverage data to the UCSC browser and examined the read coverage near the offending region (chr12:69154464-69154707) and found that there two were large spikes in coverage in two subsequent regions on chr12, roughly at positions 69158856...69160012 and 69360961...69361955. So, I added the following three lines to the top of my Masks.gtf file, to see if I could force these regions to be excluded.
chr12 BEH exon 69155265 69159154 . + . gene_id "beh0"; transcript_id "beh0_region";
chr12 BEH exon 69158856 69160012 . + . gene_id "beh1"; transcript_id "beh1_region";
chr12 BEH exon 69360961 69361955 . + . gene_id "beh2"; transcript_id "beh2_region";
This did not have any noticeable effect; the program hangs in the exact same position, with the same messages.
The last thing I tried, was to try to make the program skip all 'bundles' with large numbers of aligned reads, by using the "--max-bundle-frags" flag. I tried the following scenarios, all with no noticeable effect:
--max-bundle-frags 10000
--max-bundle-frags 1000
--max-bundle-frags 100
--max-bundle-frags 1000
--max-bundle-frags 100
In fact, when I examined the output using the -v flag, cufflinks gave no indication that it was skipping bundles with out-of-range read counts. E.g. the output contained messages such as the following even when --max-bundle-frags was equal to 100:
Inspecting bundle chr12:67510720-67511387 with 20394 reads
All of the above were tried on two different versions of cufflinks (v2.0.2 and v1.3.0) on a 64 bit linux computer.
Does anyone have any other advice for me? It seems like cufflinks simply won't work on my data set.
Comment