Building Hypotheses and Searching Databases

1
Building Hypotheses and Searching Databases
2
Two ways of creating an hypothesis

• Automatically create a chemical feature-based
  hypothesis from a set of compounds with respect
  to a type of activity.
• Build an hypothesis by assembling substructures
  and chemical functions and specifying the
  geometric constraints between them.

3
Components of an Hypothesis

• Assemble the components from known data such as
  the atomic coordinates available from X-ray
  crystallographic data.
• Express the characteristics of the hypothesis as
  a collection of particular chemical substructures
  or a collection of chemical functions such as
  hydrogen bond donors and hydrophilic groups, or a
  combination of substructures and chemical
  functions.

4
Chemical Substructures Available

• The feature dictionary contains a large library
  of chemical functional groups such as primary,
  secondary, tertiary amines, hydroxyl, carbonyl,
  acridyl, acetoamido, 1-beta-glucopyranosyl, amino
  acids etc, etc.

5
Chemical Functions Available

• The chemical functions available include HB
  ACCEPTOR, HB ACCEPTOR lipid, HB DONOR,
  HYDROPHOBIC, HYDROPHOBIC aliphatic, HYDROPHOBIC
  aromatic, NEG CHARGE, NEG IONIZABLE, POS CHARGE,
  POS IONIZABLE, RING AROMATIC

6
Chemical Functions Available

• The distances, angles, and/or torsions between
  items in an hypothesis, the preferred location of
  a chemical feature, and a range of elements per
  atom position may be specified within the
  hypothesis or a substructure of the hypothesis.
• Excluded volumes may also be specified.

7
Using the hypothesis?

• Having built the hypothesis databases may then be
  searched with the hypothesis to find compounds
  within the databases that match the hypothesis.

8
Building a Substructure Hypotheses and Searching
Databases
9
Hydrogen count set to anything
10
Specifying atom range per atom position
11
Specifying atom range per atom position
12
Searching a Database with an Hypothesis

• Once the hypothesis has been designed and built
  it may be used to search a database for compounds
  that contain the defined features.

13
Results of Database search. Default 300 hits.
14
Hit example 1
15
Hit example 1
16
Compound data
17
Compound sort. Handling large datasets.
18
Compound property report.
19
Managing Databases Coping with extremely large
numbers.

• Are scientists going to manage in the new world
  without an in depth knowledge of mathematics and
  statistics?
• Are we training scientists to cope with tools
  such as, cluster analysis, discriminate analysis,
  cross validation techniques, neural networks,
  Fourier transformations etc, etc.
• Do we need courses in the design, building and
  interrogation of databases.

20
Building a Feature Based Hypotheses and Searching
Databases
21
Setting distance constraints.
22
Setting distance constraints.
23
Hybrid hypothesis of chemical functions and
fragments.
24
Using the generic b-adrenergic agonist to search
a database.
25
Databases?

• Global structural databases e.g. CCSD
• Structure specific databases.
• Therapeutic area based databases.
• Multi-conformer databases.
• Composite databases that encapsulate the
  information of multi-gigabyte files.
• QSAR based databases.
• Commercially available versus problem specific
  databases.