Classification of G-protein coupled receptors by alignment-independent extraction of principal chemical properties of primary amino acid sequences.

摘要: We have developed an alignment-independent method for classification of G-protein coupled receptors (GPCRs) according to the principal chemical properties their amino acid sequences. The relies on a multivariate approach where primary sequences are translated into vectors based physicochemical acids and transformation data uniform matrix by applying modified autocross-covariance transform. application component analysis set 929 class A GPCRs showed clear separation major classes GPCRs. partial least squares projection latent structures created highly valid model (cross-validated correlation coefficient, Q2 = 0.895) that gave unambiguous in training ligand binding class. was further validated external prediction 535 novel not included set. Of latter, only 14 sequences, confined rapidly expanding GPCR classes, were mispredicted. Moreover, 90 orphan out 165 tentatively identified could be used assess importance every single protein contributions explaining family membership. It then revealed all unaligned contributed classifications, albeit varying extent; most important being those also determined conserved using traditional alignment-based methods.