We have screened a list of NCBI RefSeq genes for peptides which are strong hits to hidden Markov models built for Class A, Class B, Class C, and Frizzled/Smoothened GPCRs.

Each screening currently includes, for each genomic peptide sequence:

• HMM E-value (measure of statistical significance of the hit). The strongest hits have the lowest E-values.

• A link to a page containing subfamily predictions for the sequence made by a library of support vector machines trained on data from the GPCRDB information system. The subfamily predictions pages also contain direct links to information in GPCRDB. The predictions for proteins with an NP accession prefix also are linked to the August 6, 2001 assembly in the U.C. Santa Cruz Human Genome Browser.

• A BLASTme button. This button will produce a list of the top ten NCBI-BLAST hits against the non-redundant protein database nr. The list includes alignments and a graphical representation of the aligned hits.