GeneFormatics: A fuzzy feeling for function

The need to sift the wealth of gene sequence information flowing from plant and animal sequencing efforts has spawned a number of companies bent on finding the genes relevant to disease or essential plant processes. Rather than rely on cell-based assay methods to detect a gene's relevance, however, these efforts are now entering the realm of the virtual. GeneFormatics Inc. is developing computational algorithms that predict a gene's function based on both its sequence and its three-dimensional structure.

Most bioinformatics departments in pharmaceutical and agricultural biotech companies already apply algorithms to sequence information to predict function, focusing on either sequence alignment or sequence motifs, said Jacquelyn Fetrow, chief scientific officer. Although powerful, she said, these techniques do not take into account the structure that the protein takes when functional...