Skip to main content
. 2020 Nov 16;16(11):e1009049. doi: 10.1371/journal.pgen.1009049

Fig 5. Sub-linear scaling of IMPUTE5.

Fig 5

(A) Time per sample spent to impute a marker in a 10Mb region for reference panel size 10K, 100K and 1M samples, when imputing 10, 100 and 1,000 target samples. The vertical axis is on a log scale. (B) Mean number of copying states selected for the simulated reference panels. (C) Mean number of copying states selected for the downsampled HRC reference panels containing 1,000 to 30,000 thousands samples. Time and number of conditioning states are obtained using IMPUTE5 neighbours select and L = 4.