Tophat problem: failing reads alignment

[Annibal]

Hi all,
i'm dealing for the first time with Rna-seq data and i'm training performing some exercises running tophat on "wgEncodeCshlLongRnaSeqK562CellTotalFastqRep1.fastq" reads data from ENCODE project.
I'v issued the cocmmand "tophat -p 24 -G genes.gtf -o K562_1 hg19 reads.fastq" where genes.gtf is the transcript annotation file (hg19) from Illumina, hg19 is the reference genome (bowtie index) and reads.fastq is the single end reads file mentioned above (152 in length)
After 15 hrs i have in the output directory a very short .bam file (around 300 Kbyte).
Looking at the log files i find:

bowtie.left_kept_reads.fixmap.log:
::::::::::::::
# reads processed: 77046522
# reads with at least one reported alignment: 82 (0.00%)
# reads that failed to align: 77046440 (100.00%)
Reported 82 alignments to 1 output stream(s)

::::::::::::::
bowtie.left_kept_reads_seg1.fixmap.log
::::::::::::::
# reads processed: 77046440
# reads with at least one reported alignment: 17276886 (22.42%)
# reads that failed to align: 59283777 (76.95%)
# reads with alignments suppressed due to -m: 485777 (0.63%)
Reported 83222144 alignments to 1 output stream(s)
::::::::::::::
bowtie.left_kept_reads_seg2.fixmap.log
::::::::::::::
# reads processed: 77046440
# reads with at least one reported alignment: 22628110 (29.37%)
# reads that failed to align: 53761233 (69.78%)
# reads with alignments suppressed due to -m: 657097 (0.85%)
Reported 119440312 alignments to 1 output stream(s)
::::::::::::::
bowtie.left_kept_reads_seg3.fixmap.log
::::::::::::::
# reads processed: 77046440
# reads with at least one reported alignment: 18924892 (24.56%)
# reads that failed to align: 57402732 (74.50%)
# reads with alignments suppressed due to -m: 718816 (0.93%)
Reported 122913616 alignments to 1 output stream(s)
::::::::::::::
bowtie.left_kept_reads_seg4.fixmap.log
::::::::::::::
# reads processed: 77046440
# reads with at least one reported alignment: 12507630 (16.23%)
# reads that failed to align: 64207195 (83.34%)
# reads with alignments suppressed due to -m: 331615 (0.43%)
Reported 56082256 alignments to 1 output stream(s)
::::::::::::::
bowtie.left_kept_reads_seg5.fixmap.log
::::::::::::::
# reads processed: 77046440
# reads with at least one reported alignment: 19159264 (24.87%)
# reads that failed to align: 57337091 (74.42%)
# reads with alignments suppressed due to -m: 550085 (0.71%)
Reported 100864131 alignments to 1 output stream(s)
::::::::::::::
bowtie.left_kept_reads_seg6.fixmap.log
::::::::::::::
# reads processed: 77046440
# reads with at least one reported alignment: 7086758 (9.20%)
# reads that failed to align: 69837946 (90.64%)
# reads with alignments suppressed due to -m: 121736 (0.16%)
Reported 26553396 alignments to 1 output stream(s)

Moreover i find a lot (around 200) of "malformed closure" warnings in long_spanning_reads.log

Thanx a lot for any suggestions/advice.

[swaraj]

I would ask you to include the genome in your tophat run

so you new command will be

"tophat -p 24 -G genes.gtf /path/to/genome -o K562_1 hg19 reads.fastq"

The genome file name would be the common prefix for files you generate using the genome fasta file and bowtie

"bowtie-build genome.fa genome"

[Annibal]

I've included it.
As mentioned above it is "hg19 ". This are the .ebwt index and bowtie correctly build the hg19.fa reference file.

I've tried the same procedure but using paired end fastq reads (2x76) from different Rna-seq (wgEncodeCshlLongRnaSeqK562CellLongnonpolyaFastqRd1Rep1.fastq.gz and wgEncodeCshlLongRnaSeqK562CellLongnonpolyaFastqRd2Rep1.fastq.gz ) and it worked.
Maybe the problem is the format of the single end reads data of 152 nt?

Thanx

[Julien Roux]

Maybe the problem is the format of the single end reads data of 152 nt?

Probably this is part of the problem since by default Tophat only allows 2 mismatches on the whole read. I had a similar problem when analyzing reads of 107bp. Switching to --bowtie-n mode might help since the mismatches are counted only in the seed region (28 first bp). But still, I found no way to increase the parameter "-e" of Bowtie from Tophat command line, and I suspect it might be too restrictive for long reads.
If you find a way to improve the alignment, please please keep us informed!

[anurag.gautam]

can anybody help with this error
I tried to map Illumina paired-end RNA seq reads of Rice to reference genome.
I ran the tophat with the following command:

/opt/tophat-1.4.1.Linux_x86_64/tophat -p 4 -o output -G osa.gtf /home/anurag.gautam/03_Genomes/Oryza_sativa_Indica/bowtie/osa SRR037735_1.fastq SRR037735_2.fastq
[Mon Apr 9 18:11:39 2012] Beginning TopHat run (v1.4.1)
-----------------------------------------------
[Mon Apr 9 18:11:39 2012] Preparing output location output/
[Mon Apr 9 18:11:39 2012] Checking for Bowtie index files
[Mon Apr 9 18:11:39 2012] Checking for reference FASTA file
[Mon Apr 9 18:11:39 2012] Checking for Bowtie
Bowtie version: 0.12.7.0
[Mon Apr 9 18:11:39 2012] Checking for Samtools
Samtools Version: 0.1.16
[Mon Apr 9 18:11:39 2012] Generating SAM header for /home/anurag.gautam/03_Genomes/Oryza_sativa_Indica/bowtie/osa
format: fastq
quality scale: phred33 (default)
[Mon Apr 9 18:11:39 2012] Reading known junctions from GTF file
Warning: TopHat did not find any junctions in GTF file
[Mon Apr 9 18:11:39 2012] Preparing reads
left reads: min. length=75, count=9884891
right reads: min. length=75, count=9873028
[Mon Apr 9 18:18:36 2012] Creating transcriptome data files..
[FAILED]
Error: gtf_to_fasta returned an error.

Please help with this error.

[Annibal]

Probably this is part of the problem since by default Tophat only allows 2 mismatches on the whole read. I had a similar problem when analyzing reads of 107bp. Switching to --bowtie-n mode might help since the mismatches are counted only in the seed region (28 first bp). But still, I found no way to increase the parameter "-e" of Bowtie from Tophat command line, and I suspect it might be too restrictive for long reads.
If you find a way to improve the alignment, please please keep us informed!

Don't know if it works, i haven't reviewed the code but i suppose you can trick the program editing the tophat file since it only calls the bowtie exec...
At line 717:
if option == "--bowtie-n":
self.bowtie_alignment_option = "-n"

Just replace the "-n" with "-e your_value -n" and when you run tophat with --bowtie-n it will invoke bowtie with -e yourvalue -n

[yingzhang]

I will first check whether the gtf file is in the right format. Then I will check whether I have allocated enough memory for TopHat. My job once got killed at the exact step because it used much more memory than I specified.

can anybody help with this error
I tried to map Illumina paired-end RNA seq reads of Rice to reference genome.
I ran the tophat with the following command:

/opt/tophat-1.4.1.Linux_x86_64/tophat -p 4 -o output -G osa.gtf /home/anurag.gautam/03_Genomes/Oryza_sativa_Indica/bowtie/osa SRR037735_1.fastq SRR037735_2.fastq
[Mon Apr 9 18:11:39 2012] Beginning TopHat run (v1.4.1)
-----------------------------------------------
[Mon Apr 9 18:11:39 2012] Preparing output location output/
[Mon Apr 9 18:11:39 2012] Checking for Bowtie index files
[Mon Apr 9 18:11:39 2012] Checking for reference FASTA file
[Mon Apr 9 18:11:39 2012] Checking for Bowtie
Bowtie version: 0.12.7.0
[Mon Apr 9 18:11:39 2012] Checking for Samtools
Samtools Version: 0.1.16
[Mon Apr 9 18:11:39 2012] Generating SAM header for /home/anurag.gautam/03_Genomes/Oryza_sativa_Indica/bowtie/osa
format: fastq
quality scale: phred33 (default)
[Mon Apr 9 18:11:39 2012] Reading known junctions from GTF file
Warning: TopHat did not find any junctions in GTF file
[Mon Apr 9 18:11:39 2012] Preparing reads
left reads: min. length=75, count=9884891
right reads: min. length=75, count=9873028
[Mon Apr 9 18:18:36 2012] Creating transcriptome data files..
[FAILED]
Error: gtf_to_fasta returned an error.

Please help with this error.