Knowing WhenKnowing Whento Give Upto Give Up

Sensible Evaluationin Virtual Screeningfor Drug Discovery

Gwyn SkoneStephen Cameron and Irina

Voiculescu

Virtual ScreeningVirtual Screening•Part of the Drug Discovery ‘Pipeline’

• Identify & validate target• Prepare ligand database Filter ligands to select leads• Lead result optimization• Pre-clinical testing• Clinical trials• Approval and marketing

Virtual ScreeningVirtual Screening

How is it done?How is it done?

DockTarget molecule

Set of LigandConformations

Set of Scored Poses

ScoreTarget molecule

Ligand PoseReal

How is it done?How is it done?Dock(T, Ls) =•For each ligand LLs...

•For each possible conformation...• For each considered pose...

• Calculate Score(T, L)

N

Score(T, L) = Fi(t, l)lLatoms tTatoms i=1

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31

-500

-325

-150

0%5%

10%15%20%25%30%35%40%45%50%

Atoms evaluated

Threshold score

Percentage of poses rejected ateach stage (atom) of scoring function

evaluation, by rejection threshold

How might we improve it?How might we improve it?•Function-Based Early Rejection

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31

-500

-325

-150

0%5%

10%15%20%25%30%35%40%45%50%

Atoms evaluated

Threshold score

Percentage of poses rejected ateach stage (atom) of scoring function

evaluation, by rejection thresholdwith the use of atom ordering

How might we improve it?How might we improve it?•Prioritized Atom Evaluation

How might we improve it?How might we improve it?•Red: percentage of poses abandoned;•Blue: reduction in run time.

• Dotted lines are the unordered case.Early-Rejection Effect Statistics

0%

20%

40%

60%

80%

-500

-475

-450

-425

-400

-375

-350

-325

-300

-275

-250

-225

-200

-175

-150

-125

-100-7

5-5

0-2

5N

one

Threshold Score

Aban

done

d Po

ses

/Ti

me

Impr

ovem

ent

How might we improve it?How might we improve it?•Result Quota Consideration

•Black line: run time; Bars: results by RMSD (Å):0 Blue 1 < Green 2 < Yellow 3 <

Red 4ÅQuota-Based Rejection Thresholds

(averaging 3 runs of 200 results)

0%

10%

20%

30%

40%

50%

All 6 12 18 24 30 36 42 48 54 60 66 72 78 84 90

Minimum number of generations

Resu

lts

0

1

2

3

4

5

Mean run tim

e (m

inutes)

Is there anything more Is there anything more positive?positive?•Use Quaternions to represent

rotations

Comparison of Rotation Representations

0%

10%

20%

30%

40%

50%

Euler Angles Quaternions

Resu

lts

2

4

6

8

10

12

Run time (m

inutes)

Is there anything more Is there anything more positive?positive?•Learn From Prior Experience

•Search Pose Pre-Selection

Is there anything more Is there anything more positive?positive?•Start in Likely Places

•Automatic Pocket Detection

end.