. 2012 Sep 11;3:315. doi: 10.3389/fmicb.2012.00315

Table 2.

Statistical analysis of 454 pyrosequencing-induced errors for five plasmid antibodies.

Antibody	Length (nt)	N_Seq	N_Iden	Unnormalized			Normalized (per 100 nt)
				RMS_Mut (nt)	RMS_Ins (nt)	RMS_Del (nt)	RMS_Mut (nt)	RMS_Ins (nt)	RMS_Del (nt)
NO ERROR CORRECTION
VRC01	363	47542	289	5.0	2.2	1.7	1.38	0.61	0.47
VRC03	390	53734	12309	4.6	4.1	0.9	1.18	1.05	0.23
VRC-PG04_cog	369	43718	21281	5.0	1.7	0.5	1.36	0.46	0.14
gVRC-H3_d74	381	53147	19843	6.3	1.9	1.1	1.65	0.50	0.29
gVRC-H6_d74	399	13639	1013	6.4	2.8	1.6	1.60	0.70	0.40
WITH ERROR CORRECTION
VRC01	363	47542	334	5.8	1.9	1.2	1.60	0.52	0.33
VRC03	390	53734	12948	4.7	4.0	0.9	1.21	1.03	0.23
VRC-PG04_cog	369	43718	22021	5.1	1.6	0.4	1.38	0.43	0.11
gVRC-H3_d74	381	53147	23097	6.5	1.7	0.9	1.71	0.45	0.24
gVRC-H6_d74	399	13639	1033	6.6	2.7	1.5	1.65	0.68	0.38

Columns include antibody name, nucleotide-sequence length of antibody heavy-chain variable domain, number of 454 pyrosequencing-determined heavy-chain variable-domain sequences for this antibody, number of sequences 100% identical to the sequenced antibody heavy-chain variable-domain, root-mean-square (RMS) fluctuation of 454 pyrosequencing-induced mutations, insertions, and deletions with respect to the input antibody sequence, and their values after normalization by a length of 100 nucleotides.

As shown in the percent sequence identity matrix in Table 1 and the divergence/identity plots in Figure 1, only five antibodies can be distinguished from others using a single sequence identity cutoff. After mapping the 454 pyrosequencing-determined heavy-chain variable-domain sequences onto 10 plasmid antibodies, a single cutoff of 75% was applied to extract sequences corresponding to VRC01, VRC03, VRC-PG04_cog, gVRC-H3_d74, and gVRC-H6_d74, respectively.

RMS_Mut, RMS_Ins, and RMS_Del were calculated using the formula:

${RMS}_{X} = \sqrt{\sum_{i}^{N_{Seq}} \frac{{(X_{i} - \bar{X})}^{2}}{N_{Seq}}}$ ,

X denotes the type of sequencing error to be characterized, mutation (Mut), insertion (Ins), and deletion (Del), respectively; $\bar{X}$ denotes the averaged sequencing error; N_Seq denotes the total number of sequences within a given antibody group.

The RMS values were normalized using RMS_normalized = RMS_unnormalized/lengh_unnormalized × 100 to take into account the difference in sequence length.