Dataset Information

Metric learning for enzyme active-site search.

ABSTRACT:

Motivation

Finding functionally analogous enzymes based on the local structures of active sites is an important problem. Conventional methods use templates of local structures to search for analogous sites, but their performance depends on the selection of atoms for inclusion in the templates.

Results

The automatic selection of atoms so that site matches can be discriminated from mismatches. The algorithm provides not only good predictions, but also some insights into which atoms are important for the prediction. Our experimental results suggest that the metric learning automatically provides more effective templates than those whose atoms are selected manually.

Availability

Online software is available at http://www.net-machine.net/?kato/lpmetric1/

SUBMITTER: Kato T

PROVIDER: S-EPMC2958746 | biostudies-literature | 2010 Nov

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Metric learning for enzyme active-site search.

Kato Tsuyoshi T Nagano Nozomi N

Bioinformatics (Oxford, England) 20100923 21

<h4>Motivation</h4>Finding functionally analogous enzymes based on the local structures of active sites is an important problem. Conventional methods use templates of local structures to search for analogous sites, but their performance depends on the selection of atoms for inclusion in the templates.<h4>Results</h4>The automatic selection of atoms so that site matches can be discriminated from mismatches. The algorithm provides not only good predictions, but also some insights into which atoms ...[more]

PMID: 20870642

Similar Datasets

Project description:In order to suggest detailed mechanistic hypotheses for the formation and dehydration of a key carbinolamine intermediate in the T4 pyrimidine dimer glycosylase (T4PDG) reaction, we have investigated these reactions using steered molecular dynamics with a coupled quantum mechanics-molecular mechanics potential (QM/MM). We carried out simulations of DNA abasic site carbinolamine formation with and without a water molecule restrained to remain within the active site quantum region. We recovered potentials of mean force (PMF) from thirty replicate reaction trajectories using Jarzynski averaging. We demonstrated feasible pathways involving water, as well as those independent of water participation. The water-independent enzyme-catalyzed reaction had a bias-corrected Jarzynski-average barrier height of approximately (6.5 kcal mol(-1) (27.2 kJ mol(-1)) for the carbinolamine formation reaction and 44.5 kcal mol(-1) (186 kJ mol(-1)) for the reverse reaction at this level of representation. When the proton transfer was facilitated with an intrinsic quantum water, the barrier height was approximately 15 kcal mol(-1) (62.8 kJ mol(-1)) in the forward (formation) reaction and 19 kcal mol(-1) (79.5 kJ mol(-1)) for the reverse. In addition, two modes of unsteered (free dynamics) carbinolamine dehydration were observed: in one, the quantum water participated as an intermediate proton transfer species, and in the other, the active site protonated glutamate hydrogen was directly transferred to the carbinolamine oxygen. Water-independent unforced proton transfer from the protonated active site glutamate carboxyl to the unprotonated N-terminal amine was also observed. In summary, complex proton transfer events, some involving water intermediates, were studied in QM/MM simulations of T4PDG bound to a DNA abasic site. Imine carbinolamine formation was characterized using steered QM/MM molecular dynamics. Dehydration of the carbinolamine intermediate to form the final imine product was observed in free, unsteered, QM/MM dynamics simulations, as was unforced acid-base transfer between the active site carboxylate and the N-terminal amine.

Dataset Information

Metric learning for enzyme active-site search.

Motivation

Results

Availability

Publications

Metric learning for enzyme active-site search.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets