| Sequence alignment algorithms play a central role in functional identification of biological sequences. The computation work described here develops methods for calibrating sequence alignment parameters to biological questions of interest. In these studies, alignments serve as surrogates for biological functions, hybridization potentials, and vocal information respectively.; In the first, a design procedure was developed to select a subset of expressed sequence tags (ESTs) as probes for cDNA microarrays. The ESTs were annotated automatically using a functional reference dataset, Gene Ontology (GO). Three different indexes for ESTs were developed based on sequence similarity, sequence contiguity, and their functional annotation in order to select subsets of ESTs that will minimize sequence redundancy and characterize potential cross-hybridization while providing functionally representative probes.; In the second study, we addressed the problem raised by the first study, the potential cross-hybridization between sequences in microarray experiments, using metrics generated by sequence analyses to identify significant predictors for hybridization. We developed a systematic variable selection scheme using multiple multivariate models (multiple linear regression, regression trees, and feed-forward artificial neural network analysis) to select a good set of predictors for hybridization, and validated this approach on a DNA microarray with p450 family genes.; In the third study, returning to the field that originated many of today's alignment algorithms, we applied the sequence alignment algorithm to identify and classify cetacean vocalizations. Prior to our study, only large cetaceans, such as humpback whales, were thought to have hierarchically-structured songs. In this study, we discovered the hierarchically structured song-like vocalizations of a toothed whale, the false killer whale (Pseudorca crassidens), and quantified the similarities among recurring acoustic thematic elements using alignment algorithms.; These three studies describe the means to calibrate the sequence alignment algorithms crucial to interpreting the functional significance of different biological sequences. |
