Pharmaceutical Informatics and Computer-Aided

companyscourgeAI and Robotics

Oct 19, 2013 (3 years and 10 months ago)

85 views



Pharmaceutical Informatics and Computer
-
Aided

Drug Discovery


Sangtae Kim

Executive Director, Morgridge Institute for Research


CDS&E Distinguished Seminar Series at Rutgers


October, 10, 2011




Twin institutes under one roof

on the UW
-
Madison campus


Vision Inspired by Wisconsin Idea

Strengthen Wisconsin as world class center for research and
commercialization to improve economy and lives of citizens.



Collaboration

Spark research collaborations across the sciences that accelerate breakthrough

discoveries to improve human health


Interaction

Foster interaction between public and

private research that breaks down

barriers between researchers, labs &

scientific disciplines


Community

Develop vibrant public space on campus

that builds community and engages the

public in the sciences and humanities




At center of campus science sites

IP Portfolios

University

Healthcare

Delivery

IP Portfolios

University

Healthcare

Delivery

Outline

The
Priorities
for
the Pharma
-
Informatics Department


Create
an information highway from bio
-

discovery to delivery,
from the promise of
genomics to the fruits of personalized
medicine (population segmentation).


Systems critique of
the R&D
Pipeline.


Focus
research resources on new and
better methods at the bottlenecks in the
discovery and development of new drugs,
e.g., lead optimization.

Why pharmaceutical informatics
?

Value

(log scale)

$

$1 per mg.

$100 per kg.

10
4

Pharmaceutical

Informatics

Phase II clinical trials

Informatics’ new frontier

Pharma/Biotech R&D Timeline

CDS&E: Enabling Role

of Data

in Computer
-
Aided Drug Design


Evolution of two distinct branches of computational
biology


Molecule
wriggling
(
solving differential
equations of
biochemical physics)


Data miners (informatics)


New generation trained to do both


Limitations of each branch


Example:

Protein
Kinases
:

Major

Targets of

21
st

century

Constituents of cell signaling pathways

Phosphorylation of other proteins



Cancer, Inflammation, Diabetes, …



e.g. MAPK, CDK2, EGFR, PKA, etc.


Largest enzyme family in
the genome: 518 members
with 7 sub
-
families.

11

Big Pharma’s Kinase
Interaction Map


H
igh throughput assay, M. Fabian et al. (Ambit Biosciences)


113 k
inases
& 17 k
inase
i
nhibitors


approved
drugs, candidates in clinical trials, research
compounds.



Fabian, M.A., et al., A small molecule
-
kinase interaction map for clinical kinase inhibitors.


Nat. Biotech
.

2005,
23
(3): p. 329
-
336.

=

Gleevec™(Novartis); Iressa™(AstraZeneca); Tarceva™(Roche); Sutent™(Pfizer); Arxxant™(Lilly); … plus more

Protein Kinase Inhibitors: Selectivity



Kinases are cross reactive because of structure, fold conservation.

Inhibitory impact across sub
-
families!

Gleevec
®
, a

Cancer

drug, also effective
against

Diabetes

!!

Targeted against ABL kinase but inhibits
PDGF also.

13

Some inhibitors

(
poisons
)
bind through non
-
conserved

features.

Pattern is not aligned with
evolution

and thus not a low
hanging fruit for simpler
informatics tools.

Protein Kinase Inhibitors: Selectivity



Kinases are cross reactive because of structure, fold conservation.

Inhibitory impact across sub
-
families!

Recent (2006) advance in aligning the
pattern of reactivity across sub
-
families:

A. Fernandez &
S. Maddipati
, J. Med. Chem.

14

Partially wrapped hydrogen bonds (
dehydrons
) attract
hydrophobic groups to get completely wrapped by

the
dehydronic force

gromacs simulation package


NVT Ensemble

TIP3p water model

PME electrostatics

Nose Hoover thermostat

100 equilibrium runs


Computation
details:



Packing differences
vs
. Pharmacological differences

16

Experiments

Fabian et al.

Nat. Biotech 2005

Theory

Fernandez &
Maddipati

J. Med. Chem. 2006

Previous Example: in principle, a hand
-
off

“results”

“hand
-
off”

Simulation runs

Database of results

Implication: progress via collaboration

When Hand
-
Offs are Not Possible

“results”

Implication: education and training

Simulation runs

Informatics on the characteristics of the entire run

Dehydrons & Wrapperones™ in Pharmaceutical Informatics

Gleevec™/imatinib on the

Cover of Time Magazine 2001

High
-
Throughput
-
Computing improves anti
-
cancer drugs


Change research paradigm from “generating lead generation”
to “optimizing
l
ead
o
ptimization”!


1
st

generation drug candidates (tweaks)


2
nd

generation drug candidates (wrapperones™)

Success factors enabled by

collaboratory

environment

Distinguished Investigator:
Ariel Fernandez (Aug. 2011)



Re
-
designing

better, next generation anti
-
cancer drugs:

s
elective wrapping deduced from dehydronic patterns.

A. Fernandez at entry to H.F. DeLuca Forum

Photo taken Feb. 2011 (seminar visit)

Machine learning expert
S. Maddipati
(right)

co
-
advised by S. Kim and A. Fernandez.

Also shown: R. Nandigam now at Aspen Tech

Photo taken summer 2007

“results”

Ultimate: enable sharing of sensitive data

The Future

Societal / Regulatory factors

Closing Thoughts

1925


Harry Steenbock

Vitamin D by Irradiation