Improving functional annotation of no n-s yno no mo us s nps with information theory

R. Karchin, Libusha Kelly, A. Sali

Research output: Chapter in Book/Report/Conference proceedingConference contribution

39 Citations (Scopus)

Abstract

Automated functional annotation of nsSNPs requires that amino-acid residue changes are represented by a set of descriptive features, such as evolutionary conservation, side-chain volume change, effect on ligand-binding, and residue structural rigidity. Identifying the most informative combinations of features is critical to the success of a computational prediction method. We rank 32 features according to their mutual information with functional effects of amino-acid substitutions, as measured by in vivo assays. In addition, we use a greedy algorithm to identify a subset of highly informative features [1], The method is simple to implement and provides a quantitative measure for selecting the best predictive features given a set of features that a human expert believes to be informative. We demonstrate the usefulness of the selected highly informative features by cross-validated tests of a computational classifier, a support vector machine (SVM). The SVM's classification accuracy is highly correlated with the ranking of the input features by their mutual information. Two features describing the solvent accessibility of "wild-type" and "mutant" amino-acid residues and one evolutionary feature based on superfamily-level multiple alignments produce comparable overall accuracy and 6% fewer false positives than a 32- feature set that considers physiochemical properties of amino acids, protein electrostatics, amino-acid residue flexibility, and binding interactions.

Original languageEnglish (US)
Title of host publicationProceedings of the Pacific Symposium on Biocomputing 2005, PSB 2005
Pages397-408
Number of pages12
StatePublished - 2005
Externally publishedYes
Event10th Pacific Symposium on Biocomputing, PSB 2005 - Big Island of Hawaii, United States
Duration: Jan 4 2005Jan 8 2005

Other

Other10th Pacific Symposium on Biocomputing, PSB 2005
CountryUnited States
CityBig Island of Hawaii
Period1/4/051/8/05

Fingerprint

Information theory
Amino acids
Set theory
Rigidity
Support vector machines
Electrostatics
Assays
Conservation
Classifiers
Substitution reactions
Ligands
Proteins

ASJC Scopus subject areas

  • Computational Theory and Mathematics
  • Biomedical Engineering

Cite this

Karchin, R., Kelly, L., & Sali, A. (2005). Improving functional annotation of no n-s yno no mo us s nps with information theory. In Proceedings of the Pacific Symposium on Biocomputing 2005, PSB 2005 (pp. 397-408)

Improving functional annotation of no n-s yno no mo us s nps with information theory. / Karchin, R.; Kelly, Libusha; Sali, A.

Proceedings of the Pacific Symposium on Biocomputing 2005, PSB 2005. 2005. p. 397-408.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Karchin, R, Kelly, L & Sali, A 2005, Improving functional annotation of no n-s yno no mo us s nps with information theory. in Proceedings of the Pacific Symposium on Biocomputing 2005, PSB 2005. pp. 397-408, 10th Pacific Symposium on Biocomputing, PSB 2005, Big Island of Hawaii, United States, 1/4/05.
Karchin R, Kelly L, Sali A. Improving functional annotation of no n-s yno no mo us s nps with information theory. In Proceedings of the Pacific Symposium on Biocomputing 2005, PSB 2005. 2005. p. 397-408
Karchin, R. ; Kelly, Libusha ; Sali, A. / Improving functional annotation of no n-s yno no mo us s nps with information theory. Proceedings of the Pacific Symposium on Biocomputing 2005, PSB 2005. 2005. pp. 397-408
@inproceedings{27bf0422208547599acfdea2889770aa,
title = "Improving functional annotation of no n-s yno no mo us s nps with information theory",
abstract = "Automated functional annotation of nsSNPs requires that amino-acid residue changes are represented by a set of descriptive features, such as evolutionary conservation, side-chain volume change, effect on ligand-binding, and residue structural rigidity. Identifying the most informative combinations of features is critical to the success of a computational prediction method. We rank 32 features according to their mutual information with functional effects of amino-acid substitutions, as measured by in vivo assays. In addition, we use a greedy algorithm to identify a subset of highly informative features [1], The method is simple to implement and provides a quantitative measure for selecting the best predictive features given a set of features that a human expert believes to be informative. We demonstrate the usefulness of the selected highly informative features by cross-validated tests of a computational classifier, a support vector machine (SVM). The SVM's classification accuracy is highly correlated with the ranking of the input features by their mutual information. Two features describing the solvent accessibility of {"}wild-type{"} and {"}mutant{"} amino-acid residues and one evolutionary feature based on superfamily-level multiple alignments produce comparable overall accuracy and 6{\%} fewer false positives than a 32- feature set that considers physiochemical properties of amino acids, protein electrostatics, amino-acid residue flexibility, and binding interactions.",
author = "R. Karchin and Libusha Kelly and A. Sali",
year = "2005",
language = "English (US)",
isbn = "9812560467",
pages = "397--408",
booktitle = "Proceedings of the Pacific Symposium on Biocomputing 2005, PSB 2005",

}

TY - GEN

T1 - Improving functional annotation of no n-s yno no mo us s nps with information theory

AU - Karchin, R.

AU - Kelly, Libusha

AU - Sali, A.

PY - 2005

Y1 - 2005

N2 - Automated functional annotation of nsSNPs requires that amino-acid residue changes are represented by a set of descriptive features, such as evolutionary conservation, side-chain volume change, effect on ligand-binding, and residue structural rigidity. Identifying the most informative combinations of features is critical to the success of a computational prediction method. We rank 32 features according to their mutual information with functional effects of amino-acid substitutions, as measured by in vivo assays. In addition, we use a greedy algorithm to identify a subset of highly informative features [1], The method is simple to implement and provides a quantitative measure for selecting the best predictive features given a set of features that a human expert believes to be informative. We demonstrate the usefulness of the selected highly informative features by cross-validated tests of a computational classifier, a support vector machine (SVM). The SVM's classification accuracy is highly correlated with the ranking of the input features by their mutual information. Two features describing the solvent accessibility of "wild-type" and "mutant" amino-acid residues and one evolutionary feature based on superfamily-level multiple alignments produce comparable overall accuracy and 6% fewer false positives than a 32- feature set that considers physiochemical properties of amino acids, protein electrostatics, amino-acid residue flexibility, and binding interactions.

AB - Automated functional annotation of nsSNPs requires that amino-acid residue changes are represented by a set of descriptive features, such as evolutionary conservation, side-chain volume change, effect on ligand-binding, and residue structural rigidity. Identifying the most informative combinations of features is critical to the success of a computational prediction method. We rank 32 features according to their mutual information with functional effects of amino-acid substitutions, as measured by in vivo assays. In addition, we use a greedy algorithm to identify a subset of highly informative features [1], The method is simple to implement and provides a quantitative measure for selecting the best predictive features given a set of features that a human expert believes to be informative. We demonstrate the usefulness of the selected highly informative features by cross-validated tests of a computational classifier, a support vector machine (SVM). The SVM's classification accuracy is highly correlated with the ranking of the input features by their mutual information. Two features describing the solvent accessibility of "wild-type" and "mutant" amino-acid residues and one evolutionary feature based on superfamily-level multiple alignments produce comparable overall accuracy and 6% fewer false positives than a 32- feature set that considers physiochemical properties of amino acids, protein electrostatics, amino-acid residue flexibility, and binding interactions.

UR - http://www.scopus.com/inward/record.url?scp=15944417881&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=15944417881&partnerID=8YFLogxK

M3 - Conference contribution

C2 - 15759645

AN - SCOPUS:15944417881

SN - 9812560467

SN - 9789812560469

SP - 397

EP - 408

BT - Proceedings of the Pacific Symposium on Biocomputing 2005, PSB 2005

ER -