Fast accurate missing SNP genotype local imputation

Table 3 Average imputation accuracies on the two cattle datasets

Methods:	fPH	NN	WNN	SVM	NeuN	MC	BL
hd-0.5%	.9151	.8701	.8748	.8538	.6518	.8012	.5449
hd-1%	.9209	.8695	.8707	.8563	.6495	.7991	.5432
hd-2%	.9174	.8624	.8599	.8484	.6437	.7933	.5412
hd-5%	.9123	.8482	.8436	.8362	.6336	.7817	.5379
hd-10%	.8968	.8232	.8344	.8057	.6204	.7771	.5378
hd-20%	.8831	.7951	.8032	.7782	.6072	.7548	.5352
ld-0.5%	.9643	.8618	.8705	.8765	.6817	.7269	.6563
ld-1%	.9627	.8601	.8674	.8732	.6799	.7270	.6527
ld-2%	.9616	.8571	.8636	.8704	.6796	.7265	.6552
ld-5%	.9598	.8480	.8468	.8577	.6765	.7230	.6532
ld-10%	.9566	.8328	.8209	.8395	.6740	.7169	.6542
ld-20%	.9492	.7951	.7686	.8029	.6680	.7110	.6528

Average imputation accuracies on the two cattle datasets, the high-density one is 700K and the low-density one is 60K. At each missing rate, the highest accuracy is in bold. ‘fPH, NeuN’ stand for ‘fastPHASE, NeuralNet’, respectively.