Empirical estimation of sequencing error rates using smoothing splines

BMC Bioinformatics

Table 2 Median error rates in mutation screening data using shadow linear regression and smoothing spline approaches^a

Samples	Expected ER	SRER	SRER Bias	EER_CS	EER_CS Bias	EER_RS	EER_RS Bias
SRR032565	0.1991	0.0493	0.1498	0.1080	0.0911	0.1082	0.0910
SRR032566	0.0734	0.0412	0.0323	0.0705	0.0029	0.0705	0.0029
SRR032567	0.2003	0.0542	0.1461	0.1111	0.0892	0.1111	0.0893
SRR032568	0.2040	0.0437	0.1603	0.1057	0.0984	0.1072	0.0968
SRR032569	0.1598	0.0509	0.1089	0.1018	0.0580	0.1014	0.0585
SRR032570	0.0985	0.0641	0.0345	0.0959	0.0026	0.0954	0.0032
SRR032571	0.1236	0.0575	0.0661	0.1406	0.0170	0.1406	0.0170
SRR032572	0.1495	0.0530	0.0965	0.1181	0.0314	0.1173	0.0323
SRR032573	0.1779	0.0912	0.0867	0.1518	0.0261	0.1506	0.0273
SRR032574	0.0986	0.0384	0.0602	0.0626	0.0361	0.0618	0.0368
SRR032575	0.1169	0.0839	0.0330	0.1228	0.0059	0.1227	0.0058
SRR032576	0.1554	0.0945	0.0609	0.1887	0.0333	0.1880	0.0326
SRR032577	0.1052	0.0694	0.0359	0.1054	0.0002	0.1055	0.0002
SRR032578	0.1143	0.0448	0.0695	0.1077	0.0067	0.1076	0.0067
SRR032580	0.0870	0.0619	0.0251	0.1169	0.0298	0.1163	0.0292
SRR032581	0.0770	0.0347	0.0424	0.0752	0.0018	0.0751	0.0019
SRR032582	0.0400	0.0041	0.0359	0.0309	0.0091	0.0310	0.0090
SRR032583	0.1224	0.0707	0.0517	0.1279	0.0055	0.1280	0.0056
SRR032584	0.1290	0.0540	0.0750	0.1052	0.0238	0.1032	0.0259
SRR032586	0.0445	0.0102	0.0343	0.0287	0.0158	0.0287	0.0159
SRR032587	0.1486	0.0786	0.0700	0.1562	0.0076	0.1552	0.0066
SRR032588	0.1240	0.0470	0.0770	0.1024	0.0216	0.1021	0.0220
SRR033543	0.1151	0.0587	0.0564	0.0828	0.0323	0.0832	0.0318
SRR033544	0.1267	0.0524	0.0743	0.1086	0.0181	0.1085	0.0182

^aBased on 1000 replicates. The frequency-based simulation approach was applied. For each replicate, we considered the top 1000 reads with the highest frequencies as the error-free reads and generated 1000 pairs of error-free read counts and shadow counts
Expected ER expected error rate in simulation studies
SRER error rate estimated using shadow regression
SRER Bias the absolute value of the difference between SRER and Expected ER
EER_CS empirical error rate estimated using cubic smoothing spline
EER_CS Bias the absolute value of the difference between EER_CS and Expected ER
EER_RS empirical error rate estimated using robust smoothing spline
EER_RS Bias the absolute value of the difference between EER_RS and Expected ER

ISSN: 1471-2105