Alignment problems (Tophat)

[rakscalloni]

Hi!

I am proceeding my samples alignment using Tophat2 (below, an example of the command used).
tophat2 -G Homo_sapiens.GRCh37.75.gtf -p 5 -o MSCd0Adip-11 Hsa_GRCh37_75 SRR490216.fastq

When the alignmente is finished, the file "align_summary.txt" shows the report below. I have more than 60,000 reads, but only 601 mapped. Does anybody knows why is this happening?
Why my reads are not mapping? How can I fix this?


Reads:
Input: 69492
Mapped: 601 ( 0.9% of input)
of these: 53 ( 8.8%) have multiple alignments (0 have >20)
0.9% overall read alignment rate.

Thank you!!

[Mchicken]

Hey rakscalloni,

how do the alignments of your 601 reads look like. Are there spliced alignments? Do the alignments all show 100% sequence similarity?

Since I`m not familiar with Tophat2 i can only give a weak recommendation, which might help at least a little bit:

Did you do quality trimming and adapter removal? Tophat2 uses the end-to-end alignment of bowtie2 which has 2 mismatches per default. If i remember correct, the end-to-end alignment allows no clippings (neither soft nor hard) and therefor contaminated or low quality ends will be considered in the alignment and increase the number of mismatches.

You could try to map your reads with bowtie2 (i know it is not for datasets that contain spliced transcripts) and use the local alignment method and then check the output and look if there occur multiple clippings at the end of your reads.

I do not think that this is the key to your problem but might help at least a little bit.

Mchicken

[GenoMax]

Have you done any QC on the reads to see if they are clean (no adapters etc)? Have you tried to take a few reads and blasted them against Hg19 to make sure they are what you are expecting them to be (i.e. human)?