Recovery of divergent avian bornaviruses from cases of proventricular dilatation disease: Identification of a candidate etiologic agent

Kistler, Amy L; Gancz, Ady; Clubb, Susan; Skewes-Cox, Peter; Fischer, Kael; Sorber, Katherine; Chiu, Charles Y; Lublin, Avishai; Mechani, Sara; Farnoushi, Yigal; Greninger, Alexander; Wen, Christopher C; Karlene, Scott B; Ganem, Don; DeRisi, Joseph L

Reads sharing 100% identity to each other or the Solexa amplification primers were filtered, reducing our initial set of 1.4 million reads to a working set of 600,000 unique reads. In order to quickly assess the homology of this set of reads to different sequence databases, we employed an iterative strategy using ELAND (Efficient Local Alignment of Nucleotide Data) and BLAST analyses. To filter reads from our analysis potentially derived from psittacine host tissue, the working set of reads were aligned to a database of all Aves sequences from NCBI (n = 918,511) using ELAND, which tolerates no more than 2 base mismatches, and discards both low quality reads and reads with low sequence complexity. Reads that did not align to the Aves database by ELAND analysis were next re-aligned to the Aves database for high stringency blastn analysis (e = 10 -7 , word size = 11), followed by progressively lower stringencies (down to e = 10-2, word size = 8), corresponding to reads containing only 22 nucleotide identities to sequences in the Aves database. To identify reads with some homology to Bornaviridae sequences in the resulting set of 322,790 host-filtered reads, we re-implemented the ELAND/iterative blastn analysis strategy (down to ≥ 15 nucleotides identity) using a database of all NCBI BDV sequences (n = 207) augmented by our previously recovered ABV sequences (n = 5). An additional iterative tblastx analysis was incorporated to capture distantly related reads that shared similarity to the known BDV sequences only at the level of predicted amino acid sequence (down to ≥ 6 amino acid identity).