Getting just a few alignments with Tophat2

[amarth]

I made an alignment for a 1.1GB file, the RNA-seq reads were in fastq (Illumina) format. The reference was in fasta, so i decided to convert the large file to fasta...

Later i made the bowtie libraries, and then i started. When tophat finished the work, i looked to the output folder, and i saw that the sequences aligned were a 28MB *.bam file, on the other hand, the sequences *rejected* were almost a 350MB *.bam file... I deduce that the file size is proportional to the sequences amount.

So, i'm a rookie on Bioinformatics, and my question is: is it normal to have such file sizes on both files, or i'm just doing it wrong?

[chadn737]

1) Do not convert the reads to fasta. That is unnecessary, Tophat takes fastq files....your references will always be in fasta, this is not an issue.

2) It sounds like you did have a lot of unmapped reads, but its impossible to diagnose the issue from what you have described. Try providing some information like, the exact commands you ran Tophat with or some examples of unmapped reads.

[amarth]

1) Thanks for the advice,

2) Could it be the seq Quality?









The Tophat2 command i ran was:

tophat2 --sequence-length 100 --max-insertion-length 3 --max-deletion-length 3 reference index3.fasta

so much thanks

[EGrassi]

What does samtools flagstat on the accepted_hits.bam and rejected looks like?

Just to mention the fact: I had some paired data with fastqc results similar to yours and quality trimming lead me to nothing, the problem was about the -r/--mate-std-dev parameters (I had paired data): it seems that in some cases tophat really needs them (-r 300 --mate-std-dev 50 gave me a 60% percentage of properly paired reads against the 5% without them...I will never trust the FAQ/manual again ).

[kmcarr]

2) Could it be the seq Quality?

The quality of your reads looks fine.

The Tophat2 command i ran was:

tophat2 --sequence-length 100 --max-insertion-length 3 --max-deletion-length 3 reference index3.fasta

so much thanks

There is no '--sequence-length' option for tophat. Did you mean '--segment-length'? If so 100 (presumably the full length of your read) is not an appropriate setting. The default value for --segment-length (25) is appropriate for most cases.

To diagnose the problem start using only default options and then work out from there.