Tophat versus Maq

[sbmontgomery]

I was looking generally at tophat performance for paired end reads versus Maq. In a 54x2 design from a GA2, using generally the default settings for each, it looks like Maq maps more reads and a higher fraction them are properly paired (have a SAM flag of 0x0002).

MAQ had 19.4 million of 29.1 million reads properly paired
TopHat had 6.0 million of 18.1 million reads properly paired.

Seems like a dramatic difference out of the box. Has anyone else seen this type of difference?

Cheers, Stephen

[dukevn]

I was looking generally at tophat performance for paired end reads versus Maq. In a 54x2 design from a GA2, using generally the default settings for each, it looks like Maq maps more reads and a higher fraction them are properly paired (have a SAM flag of 0x0002).

MAQ had 19.4 million of 29.1 million reads properly paired
TopHat had 6.0 million of 18.1 million reads properly paired.

Seems like a dramatic difference out of the box. Has anyone else seen this type of difference?

Cheers, Stephen

Remember that TopHat is for finding junctions only. The alignment part is done by bowtie, so the comparison is really between Bowtie and MAQ (which could be done by somebody already I guess).

D.