Previous Page  55 / 161 Next Page
Information
Show Menu
Previous Page 55 / 161 Next Page
Page Background

Conformational Ensembles from Experimental Data

and Computer Simulations

Poster Abstracts

51 

18-POS

Board 18

The Use of Conformational Ensembles in Virtual Screens

Tyler Day

1

, Brandi Hudson

2

, Levi Pierce

3

, Goran Krilov

1

, Jeremy Wilbur

3

, Jonathan Weiss

3

,

James Fraser

2

.

1

Schrodinger, Inc, New York, NY, USA,

3

Relay Therapeutics, Cambridge, MA,

USA.

2

University of California, San Francisco, San Francisco, CA, USA,

Structure-based virtual screening is a well-established protocol for identifying the initial

chemical matter of a drug discovery program. However, due to the inherent hardness of

structure-based scoring functions, the nature of the chemical matter identified by virtual screens

is strongly influenced by the protein structure employed. Conventional crystallographic

refinement produces just a single protein conformation consistent with a particular data set. In

some cases, an additional set of alternate coordinates for a small number of residues can be

modeled into high resolution data. This lack of conformational diversity can lead directly to a

similar lack of diversity in the chemical matter found during a virtual screen. To overcome this

limitation, we show how using multiple crystal structures within a single virtual screen can lead

to a significant improvement in not only the number of hits identified, but also their chemical

diversity. As an alternative approach, since multiple crystal structures are not always available,

we have constructed conformational ensembles from individual crystallographic data sets. We

find that this approach can lead to an increase in the number and diversity of hits, provided key

ensemble members are identified and included in the screen. This ensemble approach opens up

the possibility of identifying novel chemical matter that binds to lower occupancy, but still

energetically accessible, conformational states of the receptor. Such states would never be

evident in a single-structure representation of the experimental data, thereby missing the

opportunity to identify this novel chemical matter.