Clarification in the usage of the parameter S

rwdavies / STITCH

STITCH - Sequencing To Imputation Through Constructing Haplotypes

http://www.nature.com/ng/journal/v48/n8/abs/ng.3594.html

GNU General Public License v3.0

73 stars 19 forks source link

Clarification in the usage of the parameter S #71

Closed ghost closed 1 year ago

ghost commented 1 year ago

Hi, I am trying to use STITCH in my work and it is not entirely clear to me in which use-case one may want to use the S parameter,and how does it affect the imputation. The documentation just says "How many sets of founders to use".

Thanks in advance!

rwdavies commented 1 year ago

Hi,

In simplest terms, setting S > 1 averages results over multiple rounds of imputation. So for example, setting S = 2 is the same thing as running STITCH twice, and averaging the genotype posteriors and dosages for the two runs. Computational complexity is proportional to S. In my experience testing it out back in the day, it provided a marginal benefit.

Hope that helps, Robbie