Viral deep sequencing needs an adaptive approach: IRMA, the iterative refinement meta-assembler

BMC Genomics

Table 2 Significance testing of variant alleles on H3 influenza mixtures

	Assembly-specific error test				Allele-specific error test				Assembly + Allele tests
Mix-in	Major change		Negligible		Major change		Negligible		Major change		Negligible
Percent	Fails	Sig.	Fails	Sig.	Fails	Sig.	Fails	Sig.	Fails	Sig.	Fails	Sig.
0 %	99.9 %	0.1 %	99.9 %	0.1 %	98.2 %	1.8 %	99.6 %	0.4 %	100 %		99.9 %	0.1 %
0.5	70.2	29.8	99.9	0.1	9.8	90.2	99.7	0.3	70.2	29.8	100	0.0
1	3.5	96.5	99.9	0.1		100	99.6	0.4	3.5	96.5	99.9	0.1
2	0.2	99.8	99.9	0.1		100	99.6	0.4	0.2	99.8	99.9	0.1
5		100	99.9	0.1		100	99.6	0.4		100	99.9	0.1
10		100	99.9	0.1		100	99.6	0.4		100	99.9	0.1
20		100	99.9	0.1		100	99.5	0.5		100	99.9	0.1
25		100	99.9	0.1		100	99.6	0.4		100	100	0.0
50		100	99.9	0.1		100	99.5	0.5		100	99.9	0.1

Negligible allele mixtures meant both unmixed parent viruses had frequencies ≥ 98 % or ≤ 1 % while major allele mixtures were defined as one pre-mixture donor virus having ≥ 98 % frequency and the other ≤ 1 % frequency. Variant negatives (negligible mixtures or 0 % mix-in) are highlighted in yellow while variant positives are in green. Cell data were omitted when counts were zero. The null hypothesis was that variants were produced by sequencer error. All tests were with respect to second-order corrected, one-sided 99.9 % binomial confidence intervals. The percentages of minor variant alleles not distinguishable from sequencer error is marked “fails” for failing to reject the null hypothesis. The percentage of variants rejecting the null hypothesis is marked “sig.” for significant and are candidates for calling single nucleotide variants

ISSN: 1471-2164