InvBFM: finding genomic inversions from high-throughput sequence data based on feature mining

BMC Genomics

Table 1 Comparison of different features

Feature Version	No. Calls	TP₀	FN	TP	FP	Precision	Recall
Features₁₅	1468	168	70	478	990	32.56%	70.59%
Features₈	1386	170	68	479	907	34.56%	71.43%
InvBFM	1379	170	68	479	900	34.73%	71.43%

Features₁₅ means the first 15 features are extracted, Features₈ means selecting 8 numeral features by the chi-square test from Features₁₅, InvBFM means the union of Features₈ and 2 features that lead to better results in practice. No.Calls: detected inversion count. TP true positive, TP₀ remove repeats of TP, FP false positive, FN false negative. The kernel of SVM is linear, with the penalty factor of 0.1 and the gamma of 20

ISSN: 1471-2164