Using Tophat with low quality Illumina Reads

[sphil]

Hello,


i know that Tophat is one of best choices to get exon-exon structures with short reads but my problem is that the sequencer got problems with their laser while detecting the nt. So all of the reads (completly 82 bp) get an "N" at Postions 54 / 55 of 82. My question ist can TooHat or even BLAT handle this?


regards

Philip

[lifeng.tian]

Hello,
So all of the reads (completly 82 bp) get an "N" at Postions 54 / 55 of 82. My question ist can TooHat or even BLAT handle this?

Philip

Philip,
I did comparison study between TopHat and BLAT recently on my 75 nt data set. I replaced certain position with N. Both TopHat and BLAT can handle this case well ( two N at position 50,51, TopHat found 164,481 junctions, ie. 1.3% less than original data set).

Lifeng

[sphil]

hey...
thanks for fast reply i will post my results in case....

edit:

i know it should be a new thread but i think its almost a 1 qustion / 1 answer post so:
can tophat handle long FLX-454 reads? (avg. length of 260bp) or is blat state of the art to get exon-exon structures with long reads?

greets

philip

[lifeng.tian]

hey...

can tophat handle long FLX-454 reads? (avg. length of 260bp) or is blat state of the art to get exon-exon structures with long reads?

philip

with my illumina data (130 nt, paired), BLAT works better than TopHat, it
found 10% more known Refseq splice junctions.

It'll be interesting to compare both with your data. My bet is BLAT will win

[sphil]

hey, ok i will brief you if anything unpredictable will happen!

[ParthavJailwala]

Hello,


i know that Tophat is one of best choices to get exon-exon structures with short reads but my problem is that the sequencer got problems with their laser while detecting the nt. So all of the reads (completly 82 bp) get an "N" at Postions 54 / 55 of 82. My question ist can TooHat or even BLAT handle this?


regards

Philip

Hi Philip,
Tophat internally uses Bowtie which has a combination of parameters that you can adjust to allow it to map in the presence of 'n' mismatches ( I am assuming a 'N' at any position is treated as a mismatch by the algorithm). Some of these parameters are:

--initial-read-mismatches Reads are initially mapped, allowing up to this many mismatches in each read alignment. The default is 2.

--segment-mismatches Read segments are mapped independently, allowing up to this many mismatches in each segment alignment. The default is 2.

-m/--splice-mismatches <int> The maximum number of mismatches that may appear in the "anchor" region of a spliced alignment. The default is 0.

Hope that helps,

Thanks