Table 2.

Performance metrics for RUBRIC selective sequencing experiments including preliminary lambda DNA runs A1 through B1, mainline lambda experiments B2 through E2, and application examples F and G.

		A1	A2	B1	B1*	B2	C	D	E1	E2	E2*	F	F*	G	G*
Odd Reads/min		1644	1464	1873	2426	2185	1040	2700	1163	1465	1465	2095	2117	1560	1480
In-Threshold Reads/pore/min		826.3	899.3	296.6	371.0	390.5	558.3	243.7	155.3	203.7	205.0	99.95	104.0	297.4	292.5
Odd Reads/pore/min		8.959	8.596	10.39	11.40	11.95	6.868	13.02	6.032	7.846	7.834	9.999	9.975	12.82	13.17
In-Thresh Reads/pore/min		4.485	5.366	1.668	1.768	2.062	4.045	1.198	0.779	1.061	1.065	0.474	0.490	2.469	2.619
Average Pore Vacancy		70.2%	64.6%	65.2%	71.0%	72.0%	65.2%	77.2%	81.9%	84.0%	83.9%	98.3%	98.3%	78.2%	77.2%
Absolute Sequence Enrichmenta,b		0.578	0.759	0.135	0.972	0.987	0.949	0.888	0.756	0.940	0.939	0.413	0.422	1.128	1.149
Absolute Read Enrichmenta,c		0.580	0.782	0.134	0.949	1.021	0.926	0.797	0.761	0.902	0.901	0.449	0.456	1.010	1.055
Relative Read Enrichmenta,d		1.674	216.5	362.2	209.3	198.4	329.5	182.2	160.9	131.3	136.5	281.8	289.1	298.3	288.0
Throughput Ratioa,e		1.102	1.293	1.025	1.181	1.212	1.271	1.185	1.067	0.920	0.924	1.081	1.092	1.082	1.093
Decision Efficiencyf		63.6%	72.1%	14.7%	94.0%	99.3%	86.1%	98.8%	99.4%	99.5%	99.5%	97.6%	97.7%	99.3%	99.9%
Timeout Fractiong		44.7%	4.5%	100.0%	98.4%	96.0%	44.7%	99.8%	0.0%	0.0%	0.0%	0.0%	0.0%	97.5%	82.6%
Threshold Filterh	Sensitivity	0.988	0.985	0.973	0.979	0.929	0.987	0.892	0.799	0.952	0.951	0.282	0.263	0.972	0.972
	Specificity	0.581	0.540	0.919	0.931	0.918	0.566	0.965	0.920	0.902	0.902	0.956	0.955	0.917	0.917
	Precision	0.135	0.089	0.525	0.508	0.462	0.114	0.566	0.372	0.373	0.372	0.010	0.009	0.583	0.593
	Accuracy (MCC)	0.276	0.215	0.682	0.678	0.621	0.249	0.693	0.508	0.562	0.561	0.046	0.041	0.718	0.724
Skip/Sequence Decisionh	Sensitivity	0.996	0.992	0.915	0.915	0.859	0.991	0.884	0.930	0.902	0.904	0.961	0.965	0.804	0.795
	Specificity	0.584	0.982	0.995	0.995	0.993	0.985	0.991	0.951	0.963	0.963	0.998	0.998	0.996	0.996
	Precision	0.122	0.765	0.976	0.976	0.962	0.853	0.962	0.793	0.801	0.802	0.715	0.714	0.457	0.475
	Accuracy (MCC)	0.266	0.863	0.934	0.934	0.892	0.913	0.904	0.828	0.824	0.825	0.828	0.829	0.604	0.612
RUBRIC Overallh	Sensitivity	0.633	0.717	0.133	0.832	0.795	0.847	0.790	0.799	0.865	0.866	0.602	0.599	0.771	0.776
	Specificity	0.884	0.994	1.000	1.000	0.999	0.994	0.999	0.995	0.995	0.995	1.000	1.000	0.999	0.999
	Precision	0.122	0.765	0.976	0.976	0.962	0.853	0.962	0.793	0.801	0.802	0.715	0.714	0.457	0.475
	Accuracy (MCC)	0.240	0.734	0.355	0.899	0.871	0.844	0.870	0.791	0.829	0.830	0.656	0.654	0.593	0.607

^aNormalized with respect to even and odd total active pore times indicated in Supplementary Table S1.

^bCumulative sequence decision target read length/cumulative odd target read length.

^cSequence decision target read count/odd target read count.

^dSequence decision target/non-target read count divided by odd target/non-target read count.

^eEven sampled read count/odd sampled read count.

^f% of in-threshold reads receiving a skip or sequence decision.

^g% of undecided reads not receiving a decision due to the RUBRIC 2-second queue timeout period.

^hBinary classifier-based performance metrics are detailed in Supplementary Section S1.

*Dataset time-filtered to eliminate reads from periods of failed skipping, see Supplemental Section S3.