Table 2.

T1 thresholds

	# Reads1	# Reads2	# Alleles	T1Galan Negmult	T1Galan Simul	T1Simul VarAmplEff	T1Res1	T1Res2
ID	1186	306	5	50	50	74	62	66
102	418	326	6	62	61	92	176	64
153	329	205	6	62	61	93	111	121
161	416	382	4	38	38	77	113	80
165	252	524	5	50	50	128	240	137
190	295	418	6	62	61	138	102	139
211	285	443	4	38	38	63	155	69
215	506	829	6	62	61	64	64	59
223	410	335	5	50	50	100	88	87
225	458	383	4	38	38	58	53	52
248	239	463	6	62	61	68	67	71
252	536	549	5	50	50	73	61	65
256	409	297	5	50	50	74	117	93
266	341	342	5	50	50	86	105	57
272	317	317	5	50	50	53	53	69
281	430	324	7	75	74	120	111	177
347	171	261	6	62	61	83	96	73
430	300	288	4	38	38	62	57	43
449	545	283	4	38	38	82	91	75
492	133	1043	2	15	15	16	18	15
493	232	221	5	50	50	65	114	88
4695	361	701	6	62	61	113	142	120
4787	71	78	3	27	27	56	42	50
5092	273	535	6	62	61	98	75	111
5116	202	378	6	62	61	69	63	77
C3659	254	496	4	38	38	55	50	45
GO3120	220	456	5	50	50	203	93	314
GO3131	346	618	8	87	85	124	87	98
GO3132	511	794	4	38	38	51	54	53
GO3133	281	453	7	75	74	124	78	94
GO3134	363	678	8	87	85	124	102	342
GO3382	492	776	8	87	85	153	150	257
GO3394	482	609	3	27	27	28	25	25
GO3899	322	506	7	75	74	81	87	95
GO3922	235	359	6	62	61	75	68	136
GO3957	440	654	9	100	97	120	101	104

The threshold T1 defined as the minimum number of sequences per individual required for reliable genotyping was computed by Galan et al. [13] using a negative multinomial distribution (referred as T1_{Galan Negmult}). We approximated the threshold by a much less computationally intensive simulation approach (T1_{Galan Simul}). T1_{Galan Negmult} und T1_{Galan Simul} do not take differences in allele amplification efficiencies into account. We used a simulation approach to investigate the effect of variation in amplification efficiency between alleles on the minimum number of reads required to achieve a reliable genotyping (T1_{Simul VarAmplEff}). Assuming that we oversampled each amplicon, we estimated what would have been the minimum number of reads that would have given us the same genotypes as obtained with our 454 data for the two amplicon replicates per individual by resampling (T1_Res1, T1_Res2).