bowtie2 vs. Tophat2 RNA-Seq

tschauer · 12-18-2012, 02:05 AM

Hi,

I have PE 100 Drosophila HiSeq RNA-Seq data. I mapped the same data using Tophat2 (changed only the inner distance estimated by Cufflinks) and Bowtie2 against the genome.

I converted the SAM file to Bedgraph file using HTSeq.

Somehow in the case Tophat2 I see "flat" intronic coverage and in bowtie2 not (see attachment). Did anyone have similar issues?

The read counts per gene in the exons obtained by htseq-count do not really differ. Only in a few special cases.

Thanks,

T.

rboettcher · 12-18-2012, 03:05 AM

TopHat2 is designed for RNAseq and therefore splicing aware, whereas bowtie2 is designed for DNAseq and therefore cannot handle splice junctions.

Since your RNAseq is expected to come from exons only, bowtie2 maps to the exonic regions while TopHat2 considers the introns in between as spanned by split reads or read pairs. Therefore I assume your bedgraph visualizes the spanning reads by inserting these "flat" regions in between your exons.

tschauer · 12-18-2012, 03:45 AM

Thanks for the quick reply.

I also assumed so that these are only the "connections". In that case these graphs have to be read somewhat differently...

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
bowtie2 parameters for chip-seq	analyst	Bioinformatics	8	10-04-2013 02:20 AM
Tophat2 very slow when running over Bowtie2	jdenvir	Bioinformatics	2	02-18-2013 06:28 AM
tophat2/bowtie2 inconsistency in number of unmapped reads	manianslab	Bioinformatics	2	07-13-2012 01:56 PM
why low mapping rates for RNA-seq with tophat2	IceWater	RNA Sequencing	5	06-18-2012 12:41 PM
RNA-seq Tophat2 errors	LiamT	Bioinformatics	5	05-09-2012 02:35 AM

12-18-2012, 03:05 AM	#2
rboettcher Member Location: Berlin Join Date: Oct 2010 Posts: 71	TopHat2 is designed for RNAseq and therefore splicing aware, whereas bowtie2 is designed for DNAseq and therefore cannot handle splice junctions. Since your RNAseq is expected to come from exons only, bowtie2 maps to the exonic regions while TopHat2 considers the introns in between as spanned by split reads or read pairs. Therefore I assume your bedgraph visualizes the spanning reads by inserting these "flat" regions in between your exons.

12-18-2012, 03:45 AM	#3
tschauer Junior Member Location: Munich Join Date: Oct 2012 Posts: 6	Thanks for the quick reply. I also assumed so that these are only the "connections". In that case these graphs have to be read somewhat differently...