Comprehensive viral oligonucleotide probe design using conserved protein regions

Jabado, Omar J.; Liu, Yang; Conlan, Sean; Quan, P. Lan; Hegyi, Hédi; Lussier, Yves; Briese, Thomas; Palacios, Gustavo; Lipkin, W. I.

The Pfam database is comprised of hand curated seed protein alignments that are converted to a probabilistic representation using HMMs. These HMMs are used to search the protein database for homologues that can be added to the seed to create a comprehensive alignment (23, 24) . Pfam domains were analyzed to identify short, conserved protein regions and corresponding nucleic acid sequences. In the first step, the log-odds score for each position of the HMM built from the seed alignment was summed; lower scores were considered to indicate conservancy. The most conserved, non-overlapping 20 amino acid (aa) regions were identified. In the second step, protein alignments of all Pfam-A families were extracted and mapped to their underlying nucleotide sequences by cross reference to the EMBL records. HMM parsing modules from the BioPerl package were used. In the third step, the underlying nucleotide sequences were extracted and stored. In cases where the region contained gaps, flanking nucleotides were brought together to yield sequences of length 60. These sequences formed the basis for downstream probe design. Domain alignments in the Pfam-B were not used in probe design because they are of lower quality; also, as domain quality improves these alignments will be integrated into Pfam-A (23).