How to interpret Centred IBS kinship matrix produced by TASSEL?
0
0
Entering edit mode
2.5 years ago
rimgubaev ▴ 270

Hello everyone! When I utilize mixed linear model in TASSEL I include the kinship matrix as random effect to account for kin relationships between the individuals. The kinship matrix in TASSEL is calculated using Identity by state method (Centered IBS) which seems to me as a biologist quite reasonable since the identity by descent matrix is inappropriate in case if we do not sure about pedigree of individuals.

I tried to visualize this matrix (diagonal and off-diagonal elements) and as a result I got this histogram:

And this one confused me great: I tried to figure out how it is calculated by reading this paper cited by TASSEL. Since I am not mathematician the only thing I can understand after reading the article and looking at my picture that "positive peak in the bimodal distribution arises from relationships between lines with the same row number, while the negative peak corresponds to relationships between lines with different row numbers."

If you are faced with the same problem or have an expertise in this question could you, please, explain, why this matrix contains negative values and what do they mean? Why there is such strange range [-0.35, 1.34] not [0,1], where 0.5 correspond to siblings, 0.25 to cousins etc? Where are the closely related individuals (siblings and cousins) in this histogram?

I also tried to look at this presentation to deal with this issue, but there is nothing on negative values.

Thank you!

Kinship Centered IBS TASSEL • 1.6k views