Notes

Outline

The IUPAC Chemical Identifier Steve Stein, Steve Heller, Dmitrii Tchekhovskoi National Institute of Standards and Technology Gaithersburg, MD, USA   CAS/IUPAC Conference on Chemical Identifiers and XML for Chemistry Columbus, OH July 1, 2002

IUPAC & Chemical Identity Mission International, open standards for chemical communication Printed Media – Nomenclature Human communication Rules for structure to name conversion Digital Media – Identifier Computer communication Rules for structure to identifier conversion Freed from restrictions of ‘pronouncibility’ Freed from ring index

Chemical Identifiers Structures Connection Tables ‘Trivial’ Names Systematic Names Index Numbers

Too Many Identifiers Structure diagrams various conventions contain ‘too much’ information Connection Tables MolFiles, Smiles, ROSDAL, .. Pronounceable names IUPAC, CAS, trivial Index Numbers EINECS, FEMA, DOT, RTECS, CAS, Beilstein, USP, RTECS, EEC, RCRA, NCI, UN, USAF

What kind of Identifier is needed? Exactly one Identifier per structure Defined by algorithms Comprehensive Openly available Implemented

Requirements Different compounds have different identifiers All distinguishing structural information is included

Requirements One compound has only one identifier No unnecessary information is included

IChI Scope First Version Discrete, covalently bonded compounds foundation for other classes Isotopes Stereochemistry sp3 - tetrahedral Z/E - double bond Tautomers

3 Steps to IChI ‘Normalize’ Input Structure Implement chemical rules ‘Canonicalize’ (label the atoms) Equivalent atoms get the same label ‘Serialize’ the Labeled Structure A unique series of bytes

NORMALIZATION

Simplifications Ignore ‘Electron Density’ Double/triple bonds, Odd-electrons, Charges Still use for Z/E stereo perception Free Rotation Around Single Bonds Divide IChI into Layers

Ignore Electron Density Not required for compound identification Distinguishes ‘excited states’ Avoids problems Delocalization, aromaticity, zwitterions, …

Slide 13

Slide 14

Slide 15

Assume Free Rotation Around Single Bonds

LAYERS

Divide into ‘Layers’ Separate ‘Name’ into Fragments by Connectivity Isotopes Stereochemistry Tautomerism

Basic Layer Simple Connectivity Just atoms and their neighbors Ignore everything else Robust basic identifier

Isotopes

Stereochemistry Double Bond (Z/E) from coordinates or bonding Tetrahedral (sp3) ‘in/out’ bonds or x,y,z coordinates

Varieties of Double Bond Isomers

sp3 (tetrahedral) stereoisomers

Identify Stereogenic Centers Speed up processing Helpful for chemists

Basic Tautomer Layer

Tautomers

Electronic Layer Simply Store Net Charge

Slide 28

IChI Output 9 possible fields Basic ## Isotopic ## Stereo ## Stereo ## Tautomeric ## Isotopic ## Stereo ## Stereo ## Electronic ##

Possible Output Format      Example: Benzene Represent atoms as sequence number in formula      C6H6   =  C  C  C  C  C  C  H  H  H  H  H  H      tags          1  2   3  4   5   6   7  8   9 10 11 12      Basic Layer:      <basic>C6H6 1-2-7 2-3-8 3-4-9 4-5-10 5-6-11 7-12</basic>

Other Output Information Only For user verification Label true stereogenic atoms Identify equivalent atoms Warnings Unusual valences Unrecognized input ‘Reversibility’ Information Coordinates Electron density Positions of double/triple bonds, charges, odd electrons

Slide 32

Slide 33

Slide 34

Slide 35

Two Fundamental Problems Chemists Different ways to represent the same thing Different definitions of tautomerism Different guesses Chemicals Structures can depend on conditions Tautomers can depend on conditions

Slide 37

Drawing Standard Needed? Bond/No bond

Allow Full ‘Reversibility’? Coordinates Structure display Original bonds and charges For display and future use Original numbering Map to input data

IChI – What can’t it do? Discover that two structures with different connectivity represent the same compound Unless they are tautomers Predict potential for Z/E isomerism in open shell conjugated networks Cannot predict rotational barriers Fix improperly entered data Guarantees wrong IChI for bad data Properly treat non-covalent bonding Coordinate bonds Represent ‘exotic’ stereochemistry

Version I Implement All Normalization Rules – 12/02 Test against available data sets – 3/03 Final External Testing and Refinement – 7/03 Documentation, source, executable – 12/03? Open discussions ichi-l@list.rsc.org

Future Extensions Organometallics Coordinate bonds Other Stereo Forms Non-atom centered Conformations Hydrogen Bonding Polymers/Macromolecules Compound Classes Markush structures