Molecular Rift: Virtual Reality for Drug Designers

Interactive Drug Design in Virtual Reality
Ching-Man Tse, Hongjian Li, Kwong-Sak Leung, Kin-Hong Lee, Man-Hon Wong
Department of Computer Science and Engineering, Chinese University of Hong Kong
{cmtse,hjli,ksleung,khlee,mhwong}@cse.cuhk.edu.hk
Abstract—Discovering new drugs for emerging diseases has
been a challenging task. There are numerous drug design
techniques including fragment-based and diversity-oriented
methods but their accuracies and efficiencies are low. By
incorporating visualisation, biomedical experts can interact
with the process to produce drug-like ligands more efficiently.
The paper presents an interactive drug design algorithm which
generates lead candidates against a protein. A set of drug
candidates, created by an inhouse fragment-based method and
docked on the target protein, are visualised in the virtual reality
settings. Biomedical experts can investigate and select some of
the ligands for further processing, aided with distance and
bonding information. It also assists the user to drag and rotate
the ligand to the binding site they find suitable. The algorithm
runs iteratively and improves the quality of lead candidates
every step. The paper compares the quality of resulting ligands
between interactive and automatic approaches.
Keywords-Interactive Drug Design; Virtual Reality;
I. INTRODUCTION
The momentum in searching for compounds of medical
uses continues to grow as diseases are more difficult to cure
owing to drug resistance. Out of the possible configurations
of the molecules which may have medical purposes [1], only
a minority of the molecules were synthesised and exploited.
It becomes crucial to design new molecules which can be
of medical values but not explored. Computational methods
have been employed because of low cost and high efficiency.
There are numerous computational techniques, mainly
fragment-based and diversity-oriented, to generate drug candidates
for wet-lab experiments [2], [3]. Fragment-based
method constructs a library of structurally diverse small
molecules that could become fragments of active drugs
[4], [5]. The drug candidate starts with low affinity for
the target in which is systematically altered and enlarged,
generating high affinity, drug-like lead compound. Diversityoriented
method produces a library of structurally diverse
drug-like compounds, usually from common intermediates
[6], [7]. The compounds are then screened and high affinity
candidates are optimised for further analysis. The accuracy
of both strategies rely on the screening procedure, which
could be a docking program in a recent approach [8].
In addressing the accuracy issues, there are interactive
approaches which enable optimisation to the compound using
the user’s knowledge about structure-activity relationship
[9]. The algorithm visualises a set of drug candidates for the
user to choose from and generate new candidates based on
their choice using evolutionary algorithm. It overcomes the
difficulties in creating the fitness function to assess drug
design. However, it considers solely on the structure of the
drug candidates while the target protein is often known
especially for pharmaceutical companies.
We propose an interactive algorithm for drug design with
a known target protein. Visualisation plays an important
role in displaying ligand-protein structure in 3D. The user
can then investigate the structure and determine whether a
drug candidate is viable. While there are good visualisers
such as JMol and ligand editors by MolSoft, manipulating a
drug candidate around requires adequate depth information.
Virtual reality enables immersive experience to the user
where conventional visualisation techniques cannot. The
program generates a set of drug candidates by evolutionary
algorithm and refined by chemical rules and docking. The
user gives feedback by translating or rotating the candidates
and remove them if found unsuitable. Our study shows
the resulting lead compound is lighter in molecular weight
which is more readily absorbed and has comparable affinity
to the automatic method.
The detailed design and algorithm is presented in section
III. Experiments are described in section IV A discussion on
the advantages and improvements of the interactive approach
is given in section V.
II. RELATED WORK
There are many attempts in combining visualisation with
the biomedical field. Visualisation and sometimes virtual reality
are brought into place to help solve complex biomedical
problems such as protein docking and drug design.
1) Interactive Drug Design: A drug design tool Molecule
Evoluator [9] uses atom-based evolutionary approach to
explore multiple configurations of drug candidates. The tool
displays numerous configurations in 2D for the user to
choose from and modify. The evaluation however depends
completely on the user. In addition, when the target protein is
available, the tool cannot take advantage of the information
by estimating the affinity such as docking.
2) Virtual Reality in Biomedical Field: The function and
interaction of proteins depend heavily on their conformations
which are best visualised in virtual reality. The advantages
of comparative visualisation was investigated in [10]. There
is a shortcoming in their visualiser that it does not support
the standard Protein Data Bank [11] format.