SweeD output / H-scan for Selective sweeps comparison

[prakash kancherla]

Dear all,

This is my first time using SweeD to look into the selective sweeps, My genome is haploid of 22 individuals and I have a total of 938515 SNPs generated from Freebayes. I have SweeD with default parameters, 

./SweeD -name TEST -input /Volumes/sweed/input.vcf -grid 10000

could some one please explain on what basis should i select the #-grid values. my output from the sweed looks wired (or i might have given wrong parameters). please look into the screenshot of my output file from Sweed and H-scan (c++ compiler to detect selective sweep).

My main question is why does my position change into decimals and not specific to the SNPs as in H-scan (where X is position of the SNP and H is the homozigosity value for the specific SNPs)?

I highly appreciate if you could please make to understand the output of SweeD and also choosing the grid values.

Thank you

kind regards

Prakash

[Pavlos Pavlidis]

Hi,

the Prakash.

the -grid parameters is used to specify the number of points on which you would like to evaluate the likelihood score. For example if you have  a genome of 1,000,000 basepairs and you put -grid 10000 then (approximately) you will get one evaluation every 100bp.

I'd use a grid value that would allow me to get one evaluation every 1000 to 100 bp.

kind regards,

pavlos

--
You received this message because you are subscribed to the Google Groups "OmegaPlus" group.
To unsubscribe from this group and stop receiving emails from it, send an email to omegaplus+...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

--

Pavlos Pavlidis, PhD

Foundation for Research and Technology - Hellas

Institute of Molecular Biology and Biotechnology
Νikolaou Plastira 100, Vassilika Vouton
GR - 711 10, Heraklion, Crete, Greece

[prakash kancherla]

Hai Pavlos,

Thank you very much for your quick response, so according to the -grid 10000 my evaluation is 92bp (if i am right).

from the below data, how do i estimate the regions that had undergone selection, should i consider the likelihood values that are higher or lower, or should i consider the high or low values in alpha, Sorry for asking, i couldnt find any papers relating to this (nor in manual), i would be a great help if you can suggest me the right paper to understand the likelihood and alpha values. 

Position	Likelihood	Alpha
382.0000	4.465164e-08	1.200000e+03
474.4152	4.444763e+01	4.766310e-05
566.8305	5.693733e+01	4.763177e-05
659.2457	6.407383e+01	4.788439e-05
751.6610	6.889706e+01	4.778016e-05
844.0762	7.253436e+01	4.956800e-05
936.4915	7.530093e+01	4.873902e-05

63686.4429

4.455918e-08

1.200000e+03

Thank you once again.

/prakash 

[Pavlos Pavlidis]

Hi Parakash,

all correct! you should consider likelihood values. However, you will need to define a threshold value. Some times we find threshold values by doing simulations under neutrality. Check msms, ms, macs, fastsimcoal2 software. All of these can do neutral simulations.

best

pavlos