[syslm01]

Hi Folks,

I feel lucky to find this thread because I have been struggling with the same problems. After splitting the unusual FASTQ files, my TopHat results are still quite different from what reported in the recent published paper. Can you tell me where to find the provided SAM file? I want to try the the reported command line.

Thanks a lot,

Yi-Shiou

Hi,

you could find the SAM file in their suplemental material online.

[ychen]

Hi syslm01,

I look into the supplementary information page many times but didn't find any SAM file. Did I miss something very obvious or just look into a wrong place?

Thanks again,

Yi-Shiou

[syslm01]

hi ychen,

here is the link http://www.ncbi.nlm.nih.gov/geo/quer...i?acc=GSE20846
GSE20846_RAW.tar contains the SAM and gtf files

[ychen]

Hi syslm01,

Thanks so much for your help, I really appreciate it.


Yi-Shiou

[pascal]

Hi,

I splitted the fastq files with fastx into the original paired end files (positions 1-75 and 77-151) and ran TopHat with default settings. In the paper they did use:

tophat -p 8 -F 0.0 -r 50 -m 1 --no-novel-juncs -j ../../pooled_200bp_frags.juncs -o pooled_tophat2 -a 8 /fs/szasmg3/cole/ebwts/m_musculus/mm9/fast_mm9/mm9+wold_spikes s1_1.query75.txt,s2_1.query75.txt s1_2.query75.txt,s2_2.query75.txt

So the difference to my run is, that I left out "--no-novel-juncs" and "-j ../../pooled_200bp_frags.junc". Further I used the default bowtie mm9 reference.

I loaded the results into IGV and took a look at the results. The result is very different in many ways.
1. The coverage seems to be lower for almost every positions. accepted_hits.sam contains about half as much entries as the result from the paper.
2. My base phred quality varies while the quality from the paper is everywhere 40.
3. So many reads could be paired mapped (in IGV: "Pair is mapped = No")

I did run Tophat with different parameters, but the result should not be that different? Maybe I missed something when I separated fastq file into two? Does anyone know how to generate such a splice junction file?

Many thanks for any advise.

[linux4gen2]

Pascal,

Have you been able to improve the matches and got similar results as in the paper ?

[pascal]

Sorry, I have stopped working on this data set. If I continue I will let you know.