Ebola Virus Epidemiology and Evolution in Nigeria

Folarin, Onikepe A.; Ehichioya, Deborah; Schaffner, Stephen F.; Winnicki, Sarah M.; Wohl, Shirlee; Eromon, Philomena; West, Kendra L.; Gladden-Young, Adrianne; Oyejide, Nicholas E.; Matranga, Christian B.; Deme, Awa Bineta; James, Ayorinde; Tomkins-Tinch, Christopher; Onyewurunwa, Kenneth; Ladner, Jason T.; Palacios, Gustavo; Nosamiefan, Iguosadolo; Andersen, Kristian G.; Omilabu, Sunday; Park, Daniel J.; Yozwiak, Nathan L.; Nasidi, Abdusallam; Garry, Robert F.; Tomori, Oyewale; Sabeti, Pardis C.; Happi, Christian T.

We identified 17 consensus-level variants (9 synonymous, 5 nonsynonymous, 3 noncoding, all relative to the earliest EBOV sequence from the West African outbreak (accession No. KJ660346.2) in EBOV genomes from the 12 sequencingpositive Nigerian samples (Table 2 ). Variants characteristic of the LB5 (Liberia sublineage 5) [18] were shared by all Nigeria EBOV genomes. The Nigerian EBOV genomes also shared 3 variants not common in Liberia, at positions 4037, 17 016, and 18 754 (Table 2) . These variants were present in all Nigerian samples sequenced, including the index case (we note that 2 samples did not have coverage at position 18 574). Two of these variants were unique to Nigeria, and 1 variant, at position 18 754, was also seen in 2 EBOV genomes from Liberia (accession Nos. KT725314 and KT725261), suggesting a close relationship of the Nigeria clade to those samples. Two Nigerian samples had unique additional consensus variants. We also identified 31 intrahost single-nucleotide variants (iSNVs) in 5 of the 12 EBOV genomes from Nigeria (5 synonymous, 5 nonsynonymous, 5 noncoding SNPs, and 16 insertions/ deletions) (Supplementary Data 2) . We sequenced each of the 5 samples with iSNVs at least twice from replicate libraries, and iSNV calls were concordant between libraries. Eight of these iSNVs were shared by ≥2 samples, and 2 iSNVs (positions 7551 and 10 503), both found in sample E027, were also consensus variants in sample E030. The presence and number of iSNVs found correlated roughly with sample coverage; only samples with >100× coverage had >1 iSNV call that passed our basic filters.