Virmid: accurate detection of somatic mutations with sample impurity inference

Table 2 Improved mutation calling in higher coverage.

α^a	n.Answer^b	Coverage = 40 ×					Coverage = 100 ×
		n.Predict^c	n.Correct^d	Precision	Recall	F-score	n.Predict^c	n.Correct^d	Precision	Recall	F-score
70%		1999	1976	0.98	0.87	0.93	2208	2198	1.00^e	0.97	0.98
80%	2265	1551	1516	0.98	0.67	0.79	2142	2133	1.00^f	0.94	0.97
90%		638	611	0.96	0.27	0.42	1572	1545	0.98	0.68	0.81

Precision and recall values are compared between two different coverages (40 × and 100 ×) at highly contaminated data (α = 70%, 80%, and 90%). Much deeper coverage is required when severe contamination is expected.
^aα is the proportion of the control sample in the disease sample (contamination level).
^bn.Answer is the number of true somatic mutations.
^cn.Predict is the number of mutations predicted by Virmid.
^dn.Correct is the number of correct mutations in the prediction.
^eRounded up from 0.9954.
^fRounded up from 0.9958.

ISSN: 1474-760X