Question

Hi,

I have 35bp Solid colorspace sequencing data, and the actual sequences to be mapped are 20-25bp after removing the linker sequence.

I hope to find all the hits allowing no more than n mismatches (say n=3), not only the best hit.

I know there is a -M option to specify -M sensitivity,35bp. I wonder whether this setting will guarantee the best sensitivity in this case. Since my reads are only 20-25bp long, should I changed the default 4 spaced seeds to 3?

I'm new to SHRiMP, so I'd like to hear some suggestions on setting the parameters of SHRiMP to achieve the best sensitivity.

Thank you!

Answer 1

I just read the SHRiMP manual again, but I think that their explanation about -M option may not be enough to answer your question. I usually use the "seed" mode by using -s, -n, and -w and the option -M is a new feature of the version 1.3.1, which I have never tried before.

I recommend for you to use the "seed" mode--the default would be good, but please adjust the -s option if you want more sensitivity. Always fast speed compensates sensitivity and the -M option seems to exist for this purpose.

Hope my message to be helpful for your project.

Answer 2

Since my reads are only 20-25bp long, should I changed the default 4 spaced seeds to 3?

while the shrimp manual says:

• We recommend using the default 4 seeds of weight 12 in most cases.

you could try running on a smaller sample and see what happens.