Computational analysis of novel drug opportunities (CANDO)
The comprehensive solution to
characterise and treat all diseases.
We have a developed a unique computational
drug discovery platform based on fragment-based docking with dynamics,
multitargeting, and drug repurposing to discovery therapeutics with
higher efficiency, lowered cost, and increased success rates, compared
to current approaches.
We have applied this platform to evaluate how all FDA approved and
other human ingestible drugs (such as certain phenethylamines,
tryptamines, psychoactives, and dietary supplements) interact with all
protein structures (compiled from a nonredundant library of solved
protein structures as well as predicted models from various organismal
proteomes such as Homo sapiens) to identify and rank
relationships between them for all indications (diseases).
Interactions between 3733 compounds, 48,278 protein structures
encompassing 2030 indications have been determined in the first
version (v1) of the platform. The compound-proteome interaction
signatures are combined with pharmacalogical, physiological, and
cheminformatics data to predict new therapeutics through repurposing
drugs already approved for other indications. The top predictions are
verified in vitro, in vivo, and in the clinic by our
collaborators or by contract research organisations (CROs).
The project represents a comprehensive integration of our group's applied research
on therapeutic discovery, building upon basic protein and proteome
structure, function, interaction, evolution, and design research.
Funding sources include the National Institutes Health (specifically a
2010 NIH Director's
Pioneer Award), the National Science Foundation, the Kinship
Foundation, the University of Washington Technology Gap Innovation
Fund, and the Washington Research Foundation.
We are currently working with almost 30 collaborators throughout
the world to find cures for over 20 indications/diseases. See a full
list of our indications and
collaborators and some results in progress.
We have developed BINDNET, a novel
method for predicting likely binding partners for a given ligand
within a proteome of interest.
Drug discovery is protein folding with a compound.
This section is in progress. There's a lot of novelty to
this project, technically in terms of the methods used, and also in
terms of philosophy and paradigms employed (ergo, the reason for the
Director's Pioneer Award). Here are a few of them:
- Docking with protein structure + ligand dynamics.
- Automated binding site identification.
- Can be used to computationally assess new compounds from combinations of fragments (+).
- Using all the known information about current drug and drug like compounds.
- Learning from affinity measures separating entropy and enthalpy.
- Predict toxicity through nonspecific binding.
- Predict ligand-target networks.
- Fragmentation of drugs to identify pharmacophore.
- Drug comparisons to substrates and metabolites to find NCEs in the structural context of the binding site
- Drug profiles across multiple targets (not single drug per target paradigm).
- Molecular and systems level integration because of drug profile (i.e., how each compound interacts with the interactome).
- Exploiting the fact that all drug discovery thusfar has been a feature of Evolution.
- Consolidates almost all one off inhibitor discovery in one shotgun approach.
- Systems based drug discovery.
- New compounds (+) predicted to be nontoxic can be used to explore beyond the CANDO space for very intractable diseases.
- Can be used to create a system of existing and novel small molecules to manipulate living (and nonliving) systems
- If successful, it will move compbio frameworks forward unlike never before.
Ultimately the goal is personalisation to improve quality of life,
including personalised medicine. When I first came across genetics, my
dream was that each person would have their genome sequence and a
powerful computing cluster (these days, one can get a
personal supercomputer for ~$6000) where they could evaluate the
response of their proteins and proteomes (corresponding to their
specific genes and genomes) against entities in the environment, such
as bioactive chemical compounds, to improve their quality of life,
i.e., to treat and/or cure diseases as well develop vaccines. This
project is part of that dream and we're going to rigourously evaluate
whether it can come to fruition.
Everyone has a major responsibility, with some overlap. The rest of
our group also helps.
- Ram Samudrala - PIon.
- Gaurav Chopra - fragment based docking with dynamics, shotgun systems and synthetic biology, operations.
- Janez Konc - fragment based docking with dynamics.
- Jim Schuler - interactome modelling.
- Kaushik Hatti - web application design.
- Manoj Mammen - applications.
- Matt Hudson - developer.
- Ambrish Roy - in virtuale bioininformatic docking pipeline.
- Andrew Ho - personalisation, individualised webbot.
- Brian Buttrick - function prediction for docking site identification.
- Brady Bernard - all around consultant, 3dtherapeutics, commercialisation.
- Brian Buttrick - in virtuale bioinformatic docking pipeline, network comparisons.
- David Beck - all around consultant.
- Ekachai Jenwitheesuk - original developer v1.
- Geetika Sethi - pipeline management, benchmarking.
- George White - collaborations, verifications, all rounder.
- Jeremy Horst - original developer, all around consultant.
- Jeremy Li - personalisation, individualised webbot.
- Ling-Hong Hung - shotgun structural and functional biology.
- Mark Minie - writing, all rounder.
- Haychoi Taing - systems and database administrator/programmer.
- Michael Shannon - former systems administrator.
- Thomas Wood - shotgun systems and synthetic biology.
(parts large and small)
- US NIH Clinical and Translational Sciences Award (2015-2020).
- US NIH Director's Pioneer Award 7DP1OD006779 (2010-2017).
- US NSF GEMSEC (2005-2011).
- US NSF CAREER Award IIS-0448502 (2005-2010).
- US NIH F30DE017522 (2006-2010).
- The University of Washington's Technology Gap Innovation Fund (2006-2007).
- Washington Research Foundation (2006-2007).
- Puget Sound Partners in Global Health (2004-2005).
- US NIH R33 (2003-2006)
- Searle Scholar Award to Ram Samudrala (2002-2005).
- The University of Washington's Advanced Technology Initiative in Infectious Diseases (2001-2014).
These are some of the key papers that have led up to the
development of CANDO v1. See also all our
publications related to therapeutic discovery as well as a comprehensive list of
all our publications.
- Sethi G, Chopra G, Samudrala R. Multiscale
modelling of relationships between protein classes and drug behavior
across all diseases using the CANDO platform. Mini Reviews
in Medicinal Chemistry 15: 705-717, 2015.
- Minie M, Chopra G, Sethi G, Horst J, White G, Roy A, Hatti K,
Samudrala R. CANDO and
the infinite drug discovery frontier. Drug Discovery
Today 19: 1353-1363, 2014.
- Horst JA, Laurenzi A, Bernard B, Samudrala R. Computational
multitarget drug discovery. Polypharmacology
- Jenwitheesuk E, Horst JA, Rivas K, Van Voorhis WC, Samudrala
paradigms for drug discovery: Computational multitarget
screening. Trends in Pharmacological Sciences 29:
- Jenwitheesuk E, Samudrala R. Identification
of potential multitarget antimalarial drugs. Journal of
the American Medical Association 294: 1490-1491, 2005.
- Jenwitheesuk E, Samudrala R. Improved
prediction of HIV-1 protease-inhibitor binding energies by molecular
dynamics simulations. BMC Structural Biology 3: 2,
Samudrala Computational Biology Research Group ||