1 of 18. 2 of 18 introduction 3 of 18 motivation virtual synthesis for the assessment of practical...

Post on 26-Mar-2015

217 Views

Category:

Documents

2 Downloads

Preview:

Click to see full reader

TRANSCRIPT

1 of 18

2 of 18

Introduction

3 of 18

Motivation

• Virtual synthesis for the assessment of practical chemical hypothesis's.

• What is the set of all molecules available from one-pot type reactions?

• What is the set of all divergent modifications I should make to my scaffold based on starting material availability and synthetic accessibility?

• What is a simple soluble photoacid that can be obtained in two steps from a commercially available materials?

4 of 18

Can “computational processes-centric chemistry” increase the efficiency with which

human needs are met?

5 of 18

Mathematical Chemistry

• (Sub)graph isomorphism.• Analytic and constructive enumeration of

graphs (i.e. rooted trees=alkanes).

• Computational complexity of organic synthesis, big “O”, NP-completeness.

Cayley, 1875

Ncarbons 1 2 3 4 5 6 7 8 9 10 11 12 13

Nalkanes 1 1 1 2 3 5 9 18 35 75 159 357 799

6 of 18

An Atlas Of Click Chemistry

Can such an atlas be queried for quantitative answers to questions like:

• How many molecules are there? How do they cluster?

• What is the “molecular diversity” of this set? Compared to natural products? Top 100 drugs?

• How likely are we to find a new and useful molecule in this set?

• How accurate is the map?

7 of 18

Every such sequence should be in the atlas.

Sharpless and co-workers, Angew. Chemie, 2001

8 of 18

Chemically “Easy” ~ Computationally “Easy”

• Only use predictably selective outcomes (to a good chemist’s best knowledge), to avoid the ocean of red-herrings.

• Limiting pathways to < 10 steps, selective outcomes are more predictable, relevant to process and easier to compute.

• This should increase the accuracy and applicability of the atlas.

9 of 18

Distributions of Molecular Properties, Such As Diversity, Are

Computable

Statements on such properties should be qualified with numbers and labeled axis’,granted the software hasn’t been quite as accessible as it is now.

10 of 18

Why Is Now A Good Time?

• Maturation of Daylight’s SMILES specification.

• Chemaxon’s set of tools and support, free to academics. Key importance.

11 of 18

What Can We Do With “Cheap” Starting Materials and “Easy”

Reactions?

12 of 18

What We Have Now

• A skeleton specification for the set of click reaction implemented in SMIRKS.

• The EPA’s HPV list converted into SMILES, with some additional contributions from internet solicitations.

• Some plumbing that uses Chemaxon’s tools to iterate over molecules/reactions and distributes the workload in a parallel fashion over a cluster.

13 of 18

Just What Are Click Reactions Specifically?

• Reactions that employ nucleophiles: • Primary amines, monosubstituted sulfonamides, oximes,

hydrazines, hydrazides, thiols, phenols…

• Paired with appropriate electrophiles:

• Primary alkyl halides, tosylates, aldehydes, epoxides, aziridnes…

• Cycloadditions like copper catalyzed acetylene-azide.

• Other reactions such as disulfide formation.

14 of 18

We Need To Encode All Of These Reactions, Efficiently

• This problem resonates with people who work in programming languages, i.e. how do we specify a large set of graph transformations in a manner that is efficient to encode, debug, update and verify.

• Simple first step is to apply have substitution enabled SMILES.

15 of 18

lambdaSMILES

• A simple, lightweight preprocesser that makes writing out reaction specifications a little easier. Basically we get macros in SMILES. i.e.

>>N(H)(H)C{@,@@}(H)(CO)(C(=O)O)

N(H)(H)C@(H)(CO)(C(=O)O)

N(H)(H)C@@(H)(CO)(C(=O)O)

16 of 18

• Hammer out a final computational specification for the click reactions. Some compromises will undoubtedly have to be made. Think in terms of EXCLUDE, REACTIVITY and SELECTIVITY.

• Do we encode some reactions by inserting products as starting materials?

• Check the list manually to ensure that it meets the goal of enriched synthetic accessibility.

• Analyze results, toss structures to VLS.

What Needs To Be Done

17 of 18

Cool Movie

18 of 18

Future Directions

• Chemical workgroup productivity, data sharing.

• Computer science of graphical programming languages.

• Machine inference via automatic data collection (ChemCrawler).

• Optimizing global chemical products over sets of chemical processes.

• Chemical education.

19 of 18

Acknoledgements

• Dr. Peter Kuhn, Scripps-PARC

• Dr. M.G. Finn, Dr. Valarie Fokin and Dr. Barry Sharpless, Scripps

top related