Tophat does not map reads in some locations on reference genome but bowtie does

Question

[heshamgibriel]

I downloaded wheat reference genome, then I split it into two chunks as the genome was too big that bowtie required it to be split into two. Then, I indexed it with bowtie, and mapped reads with tophat. This is how I called it:

tophat -o -p 15 --min-intron-length 20 --max-intron-length 20000 --max-multihits 5 <genome> <read>

I got the bam file, but surprisingly, when I tried to visualize it with IGV, I could not find reads mapped to the genome in so many coordinates. This was a bit awkward, because it did not match what I found with samtools tview, in which I could see clear reads that are mapped on this genomic region!!!

I weakly concluded that this could be an improper mapping issue with tophat that I am not aware of. Therefore, I used bowtie, and surprisingly, I found reads mapped to the reference genome which were absent with tophat!!

Why do I get mapped reads with bowtie and not with tophat?

Do you have any suggestion or an explanation to this?

Could it have to do with calling tophat wrongly??!!

Cheers

[Carlo Yague]

Could you add the version of TopHAt and Bowtie you are using and the options you choose when you call them ?

[Devon Ryan]

Newer versions of bowtie1 and bowtie2 support larger genomes like those of wheat. This will likely solve your problem.

[heshamgibriel]

I used bowtie2 and tophat2..

This is how I called bowtie:

bowtie2 -x <genome> -U <read> -S <output>

and, this is how I called Tophat:

tophat -o -p 15 --min-intron-length 20 --max-intron-length 20000 --max-multihits 5 <genome> <read>

also, I tested it without options, and I get the same results:

tophat -o -p 15 <genome> <read>