QUICK SEARCH:   [advanced] Author: Keyword(s):
Home     Help     Feedback     Subscriptions     Archive     Search     Table of Contents

This Article Summary Full Text Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Kim, J. Articles by Carlson, J. R. Search for Related Content PubMed PubMed Citation Articles by Kim, J. Articles by Carlson, J. R.

Gene discovery by e-genetics: Drosophila odor and taste receptors

¹ Department of Molecular, Cellular and Developmental Biology, Yale University, P.O. Box 208103, New Haven, CT 06520-8103, USA
² Department of Ecology and Evolutionary Biology and Department of Statistics, Yale University, P.O. Box 208106, New Haven, CT 06520-8106, USA

View larger version (13K): [in a new window]   Fig. 1. Construction of an n-dimensional protein space that allows interpolation. Each dimension represents a tested variable. Only three dimensions are shown. Tested variables included hydropathy, polarity, pI, pKa, molecular weight, and amino acid composition. Adapted from Warr et al. (Warr et al., 2001).

View larger version (13K): [in a new window]   Fig. 2. Refined parameters useful in distinguishing GPCRs from non-GPCRs. A sliding window recognizer is used to characterize the structure of a protein. A portion of an idealized GPCR is shown. Parameters selected as being particularly useful were (1) average periodicity of the hydrophobicity function; (2) average periodicity of the polarity function; (3) variance in the periodicity of the polarity function; (4) variance in the first derivative of the polarity function; and (5) amino acid usage index. Adapted from (Warr et al., 2001); parameters are described in more detail in Kim et al. (Kim et al., 2000).

View larger version (16K): [in a new window]   Fig. 3. Setting a discriminant function to maximally separate GPCRs from non-GPCRs in protein space. In a three-dimensional space, the function appears as a plane. The function was established using the training set of 750 known GPCRs and 1000 non-GPCRs. The function is used to classify novel proteins as either GPCRs or non-GPCRs, according to which side of the plane they map.

View larger version (28K): [in a new window]   Fig. 4. Testing the algorithm. The top panel shows that the algorithm correctly identified 96% of a test set of 100 known GPCRs and produced no false positives. The middle panel shows the performance of the algorithm on 100 amino-acid stretches of GPCRs and non-GPCRs. The bottom panel shows the performance of the algorithm, following retraining, with a set of GPCRs and ion channels.