segment length in tophat

[ickou]

Hi,

I have an illumina single read sequencing with read length 36.
By default, tophat cuts segments of 25 bases.
How does tophat cut reads of 36 bases in segments of 25 bases?
Do I need to change "--segment-length" option to 18?
Can you help me to find the best parameters for single-end reads of 36 bases?

Thanks,

Ickou

[luxmare]

Hi Ickou,

As you may already know, The recommended parameters for very short reads were described in the release note of TopHat 1.3.1 as below.

For short reads (usually <45-bp), it is recommended that users decrease segment length (--segment-length) to about half the read length and segment mismatches (--segment-mismatches) to 0 or 1

Hope this helps.

Thank you,

luxmare

[mapardo]

What would the consequence of use a segment length longer than half the read length be?

[jimineep]

I am admittedly guessing this based on my (mis?)understanding of tophat but:

As I understand it, tophat will take the maps that do not map to the genome at first, and divide it into fragments of length segment-length, then try to map these. If they do map, it will then see if the other fragments from the read map (presumably somewhere else nearby).

So if the segment-length is longer than half the read length, it can't map the second fragment, because there isn't one (since you've already divided up the read into the minimum number of fragments, i.e. 1)

I have a question: If how I have described it really is the case, this means for e.g. read length 34, with segment-length 17, this means that you can only find reads which map exactly halfway accross an exon boundary? Or I've misunderstood something?

[colaneri]

segment lenght set up

In tophat manual said

--segment-mismatches Read segments are mapped independently, allowing up to this many mismatches in each segment alignment. The default is 2.
--segment-length Each read is cut up into segments, each at least this long. These segments are mapped independently. The default is 25.


In a note for the version 1.3.1 said

For short reads (usually <45-bp), it is recommended that users decrease segment length (--segment-length) to about half the read length and segment mismatches (--segment-mismatches) to 0 or 1



Should I specify these two parameters in this way?

bsub -n 4 -R "span[ptile=4]" tophat -p 4 --no-novel-juncs --segment-mismatches=1 —-segment-length=20 -G /proj/seq/data/TAIR10_Ensembl/Annotation/Archives/archive-2013-03-06-09-54-25/Genes/genes.gtf -o C_ctrl_rep1_THout_3 /proj/seq/data/TAIR10_Ensembl/Sequence/Bowtie2Index/genome C_ctrl_rep1.fastqsanger