Identifying genetic determinants of complex phenotypes from whole genome sequence data.

Finally, note that RRF was actually so fast at small chunk sizes that there was a significant overhead associated with our parallelization of the algorithm (Methods), as linear models fitted to only chunk sizes less than 200 were highly significant ( P <2.00×10 −16 ), with negative slopes (Fig. 1 b).

also in 132,142 other papers