Computer-aided synthesis design (CASD) and forward...
Transcript of Computer-aided synthesis design (CASD) and forward...
InfoChem Copyright © 2015 Dr. Valentina Eigner-Pitto ACS Fall National Meeting, Boston, August 17, 2015
Computer-aided synthesis design (CASD) and forward
reaction prediction tools for both idea generation in new
synthesis route planning and for de novo molecule design
V. Eigner-Pitto, F. Huerta*, M. Hutchings, H. Saller, P. Loew
InfoChem GmbH, Landsberger Strasse 408, Munich, 81241, Germany
*Chemnotia AB, Södertällje 151 36, Sweden
1 / 27
InfoChem Copyright © 2015 Dr. Valentina Eigner-Pitto ACS Fall National Meeting, Boston, August 17, 2015
Outline
http://www.bubblews.com/news/2372700-tips-to-be-a-professional-content-writer
Outline
• Setting the scene
o CASD and InfoChem
o New motivations
• Retrosynthesis tool: ICSYNTH
o Workflow overview
o Examples
Twistane
Oxaspiroketone
• Forward reaction prediction tool: ICFRP
o Workflow overview
o Application fields
o Interfaces and examples
• Conclusions
2 / 27
InfoChem Copyright © 2015 Dr. Valentina Eigner-Pitto ACS Fall National Meeting, Boston, August 17, 2015
Outline
http://www.bubblews.com/news/2372700-tips-to-be-a-professional-content-writer
Outline
• Setting the scene
o CASD and InfoChem
o New motivations
• Retrosynthesis tool: ICSYNTH
o Workflow overview
o Examples
Twistane
Oxaspiroketone
• Forward reaction prediction tool: ICFRP
o Workflow overview
o Application fields
o Interfaces and examples
• Conclusions
3 / 27
InfoChem Copyright © 2015 Dr. Valentina Eigner-Pitto ACS Fall National Meeting, Boston, August 17, 2015
1960 1980 2012 2014 1970 2006 2007 2011 1990 1992 1996
CASD and InfoChem Over Time
InfoChem acquires rights
for the SPRESI data
Development of ICMAP and CLASSIFY
Launch of Synthesis Tree Search (CDROM)
http://columbiascience.tumblr.com/post/53520033269/synlit-e-j-corey
Introducing Route Designer v1.0 A Peter Johnson1, Zsolt Zsoldos2,Aniko Simon2,
Darryl Reid2, James Law2, Yang Liu2, Sing Yoong
Khew2, and Howard Y. Ando3
Comprehensive synthesis
planning using multiple
reaction search algorithms Valentina Eigner-Pitto1, Josef
Eiblmaier1, Hans Kraut1, Heinz Saller1,
Peter Loew1, and Guenter Grethe2
http://neven1.typepad.com/blog/2012/12/lo
oking-for-winter-weirdness-3.html
Cooperation with AstraZeneca
Process Development
Roll-out for 50
scientists at AZ
Defined as “business
critical” in process
development department
Setting the scene 4 / 27
InfoChem Copyright © 2015 Dr. Valentina Eigner-Pitto ACS Fall National Meeting, Boston, August 17, 2015
Extensive reaction databases e.g. SPRESI (InfoChem): 4 M reactions
New Motivations
Algorithms / software Atom-atom-mapping
Reaction centre i/d
Reaction classification
Fast search engine
Computer hardware Powerful
Inexpensive
Networked
Secure
Setting the scene 5 / 27
InfoChem Copyright © 2015 Dr. Valentina Eigner-Pitto ACS Fall National Meeting, Boston, August 17, 2015
Outline
http://www.bubblews.com/news/2372700-tips-to-be-a-professional-content-writer
Outline
• Setting the scene
o CASD and InfoChem
o New motivations
• Retrosynthesis tool: ICSYNTH
o Workflow overview
o Examples
Twistane
Oxaspiroketone
• Forward reaction prediction tool: ICFRP
o Workflow overview
o Application fields
o Interfaces and examples
• Conclusions
6 / 27
InfoChem Copyright © 2015 Dr. Valentina Eigner-Pitto ACS Fall National Meeting, Boston, August 17, 2015
ICSYNTH: Workflow Overview
1. Pre-processing
2. Synthesis Planning
Automatic transform extraction
(ICMAP/CLASSIFY)
N
Remove bond 1-3
Make new single bond between atoms 3 and 5Remove bond 1-2
Make new single bond between atoms 3 and 4
1
2
3
O4
5
6
7 8
910
R111
R212
13
14
Template (2 levels):
Transform:
Make new single bond between atoms 2 and 6.
Decrease bond order of double bond 2=4 by 1.
Add group to atom 3: -OH
Remove bond between atoms 3 and 6.
Reaction database e. g.
SPRESI
proprietary databases
commercial databases
Transform library
Target molecule
N
O
O
O
O
Lookup stored examples
Precursor(s) 1
Precursor(s) 2
Precursor(s) n
...
Reaction 1:
Reaction 2:
Reaction n:
Transform x
Transform y
Transform n
Target molecule
N
O
O
O
OTransform x
Transform y
Transform n
ICSYNTH: Concept 7 / 27
InfoChem Copyright © 2015 Dr. Valentina Eigner-Pitto ACS Fall National Meeting, Boston, August 17, 2015
Examples
ICSYNTH: Examples
Publication in Organic Process Research and Development:
Route Design in the 21st Century: The ICSYNTH Software Tool as an Idea Generator for
Synthesis Prediction 2)
2) Bøgevig ‡, Federsel†, Huerta ‡, Hutchings §, Kraut §, Langer †, Low §, Oppawsky §, Rein ‡, and Saller §
†Chemical Development, AstraZeneca R&D, Macclesfield, SK10 2NA Cheshire, U.K., ‡Chemnotia AB, 151 36 Sodertalje, Sweden, §InfoChem GmbH, D-81241
Munchen, Germany
Twistane Oxaspiroketone
• Since 50 years an
interesting problem in
synthesis
• Synthesis teaching topic in many
organic chemistry courses
• Routinely used during development
• Not really interesting from the process
development point of vue
• Problem for AZ
Scale-Up Lab in
Södertälje
• Late intermediate to a series of potential
Alzheimer‘s treatements
• Medicinal Chemistry synthesis path not
suitable for scale-up
8 / 27
InfoChem Copyright © 2015 Dr. Valentina Eigner-Pitto ACS Fall National Meeting, Boston, August 17, 2015
Example: Twistane
ICSYNTH: Examples 9 / 27
InfoChem Copyright © 2015 Dr. Valentina Eigner-Pitto ACS Fall National Meeting, Boston, August 17, 2015
Example: Twistane
• 3 of the first 5 suggestions in step 1 correspond to key intermediates in published synthesis
o reassuring for new users
o gratifying for software development (in terms of e.g. algorithm and ranking capabilities)
o ...but largely irrelevent from the idea generation point of view
• New transannular C-C bond formation across decalin precursors involved in:
o ...most of the new suggested routes
o ...and in the shortest one reported in literature
o Not surprising because already historically identified as most strategic bond
• Considerations in precursor ranking within ICSYNTH
o Application of complexity measure to identify which bond(s) lead to the greatest molecular
simplification in the retro direction (bond-breaking), coupled with reactions resulting in multiple
bond formation
o Most favorable of 4 possible C-C breakages in twistane case results in decalin skeleton precursor
o Less attractive bond-making combination are more complex and lead to more complex synthesis
ICSYNTH: Examples 10 / 27
InfoChem Copyright © 2015 Dr. Valentina Eigner-Pitto ACS Fall National Meeting, Boston, August 17, 2015
Example: Oxaspiroketone
ICSYNTH: Examples
Med Chem route not suitable for scale-up:
• Yields not reproducible and < 10%
• Some reagents not optimal (TMSCN, LiHMDS)
• Tedious chromatographic separation from mixture
Additional task aspect:
• Correct stereoisomer generation, critical for
biological activity of final compound
• At least easy isolation method of desired isomer
Disconnection of bonds a, b, c and d studied:
• a and b disconnections appealing at first sight
• ...but use of non-desired reagents involved
ICSYNTH unconventional suggestion:
• Non intuitive disconnection of c in last step
11 / 27
InfoChem Copyright © 2015 Dr. Valentina Eigner-Pitto ACS Fall National Meeting, Boston, August 17, 2015
Example: Oxaspiroketone
ICSYNTH: Examples 12 / 27
InfoChem Copyright © 2015 Dr. Valentina Eigner-Pitto ACS Fall National Meeting, Boston, August 17, 2015
Example: Oxaspiroketone
ICSYNTH: Examples
• Initial conventional retrosynthetic analysis could not identify the illustrated bond c disconnection
• Although the reported precedent could not be used as such, a new route could be readily developed
Conclusion: ICSYNTH used as idea generator identifies unconventional and unusual transformations that
lead to not expected but in this case highly relevant solutions
13 / 27
InfoChem Copyright © 2015 Dr. Valentina Eigner-Pitto ACS Fall National Meeting, Boston, August 17, 2015
Outline
http://www.bubblews.com/news/2372700-tips-to-be-a-professional-content-writer
Outline
• Setting the scene
o CASD and InfoChem
o New motivations
• Retrosynthesis tool: ICSYNTH
o Workflow overview
o Examples
Twistane
Oxaspiroketone
• Forward reaction prediction tool: ICFRP
o Workflow overview
o Application fields
o Interfaces and examples
• Conclusions
14 / 27
InfoChem Copyright © 2015 Dr. Valentina Eigner-Pitto Meeting Merck - InfoChem, Darmstadt, July 22, 2015
N
B r
O
O
N
Br
O
ON
O
O
N
Br
O
ON
O
O
HBr S
O+ +
Br
OO
NN
Br
O
N N
OO
O
N
I
O
ON
O
O
N
Br
O
ON
O
O
HBr S
O+ +
N
Br
O
ON
O
O
HBr S
O+ +
N
Br
O
ON
O
O
HBr S
O+ +
N
Br
O
ON
O
O
HBr S
O+ +
N
Br
O
ON
O
O
HBr S
O+ +
N
Br
O
ON
O
O
HBr S
O+ +
N
I
O
ON
O
O
N
I
O
ON
O
O
N
I
O
ON
O
O
Target Starting material
Software Reaction Prediction Tools: ICSYNTH, ICFRP
Forward reaction prediction tool: ICFRP 15 / 60
InfoChem Copyright © 2015 Dr. Valentina Eigner-Pitto ACS Fall National Meeting, Boston, August 17, 2015
ICFRP: Workflow Overview
1. Pre-processing
2. Reaction Prediction
Automatic transform extraction
(ICMAP/CLASSIFY)
N
Remove bond 1-3
Make new single bond between atoms 3 and 5Remove bond 1-2
Make new single bond between atoms 3 and 4
1
2
3
O4
5
6
7 8
910
R111
R212
13
14
Template (2 levels):
Transform:
Make new single bond between atoms 2 and 6.
Decrease bond order of double bond 2=4 by 1.
Add group to atom 3: -OH
Remove bond between atoms 3 and 6.
Reaction database e. g.
SPRESI
proprietary databases
commercial databases
Transform library
Product(s) 1
Product(s) 2
Product(s) n
...
Lookup stored examples
Knowledge-based
compound design
ICFRP: Workflow overview
Starting material
16 / 27
InfoChem Copyright © 2015 Dr. Valentina Eigner-Pitto ACS Fall National Meeting, Boston, August 17, 2015
ICFRP Application Fields
Reactivity mapping Overview of which positions in the core
can be modified, and how
Scaffold modification Structural changes of the core
structure
Library synthesis Broad range of possible
modifications (including, e.g.,
bioisosteres) of a given
position in the molecule
Different outputs obtained by tailoring the query strategy parameters
ICFRP: Application fields 17 / 27
InfoChem Copyright © 2015 Dr. Valentina Eigner-Pitto ACS Fall National Meeting, Boston, August 17, 2015
ICFRP User Interfaces
ICFRP: Interfaces
Web-application
Knime node
Pipeline Pilot component
18 / 27
InfoChem Copyright © 2015 Dr. Valentina Eigner-Pitto ACS Fall National Meeting, Boston, August 17, 2015
Example 1: Knowledge-based Drug Design
• Proof of concept on known molecule: Diazepam
Query core
structure
ICFRP: Interfaces and examples
80 new suggested molecules
(with related synthesis reactions)
2564 known related compounds
(with associated physicochemical properties)
Reaxys substructure search
19 / 27
InfoChem Copyright © 2015 Dr. Valentina Eigner-Pitto ACS Fall National Meeting, Boston, August 17, 2015
Example 1: Knowledge-based Drug Design
ICFRP: Interfaces and examples
• Fingerprints calculated
• Identical compounds filtered off (Tanimoto = 1)
• 13 suggested products with a Tanimoto 0.7-0.9
• Physicochemical properties calculated and compared with reported ones
New “suggested”
molecules show
similar properties to
the known ones
20 / 27
InfoChem Copyright © 2015 Dr. Valentina Eigner-Pitto ACS Fall National Meeting, Boston, August 17, 2015
Example 2: Knowledge-based Drug Design
ICFRP: Interfaces and examples
• Proof of concept inspired by “de
novo design using reaction vectors”
• Example from J. Chem. Inf. Model.,
Vol. 49, No. 5, 2009, 1163-1184
Penicillin G
21 / 27
InfoChem Copyright © 2015 Dr. Valentina Eigner-Pitto ACS Fall National Meeting, Boston, August 17, 2015
Example 2: Knowledge-based Drug Design
ICFRP: Interfaces and examples
• ICFRP with medium precision
• ICFRP with high precision
22 / 27
InfoChem Copyright © 2015 Dr. Valentina Eigner-Pitto ACS Fall National Meeting, Boston, August 17, 2015
Example 3: Multistep Synthesis Starting from a Building Block
ICFRP: Interfaces and examples 23 / 27
InfoChem Copyright © 2015 Dr. Valentina Eigner-Pitto ACS Fall National Meeting, Boston, August 17, 2015
Outline
http://www.bubblews.com/news/2372700-tips-to-be-a-professional-content-writer
Outline
• Setting the scene
o CASD and InfoChem
o New motivations
• Retrosynthesis tool: ICSYNTH
o Workflow overview
o Examples
Twistane
Oxaspiroketone
• Forward reaction prediction tool: ICFRP
o Workflow overview
o Application fields
o Interfaces and examples
• Conclusions
24 / 27
InfoChem Copyright © 2015 Dr. Valentina Eigner-Pitto ACS Fall National Meeting, Boston, August 17, 2015
Conclusions
Conclusions
• ICSYNTH is used as idea generator for retrosynthetic analysis helps identifying synthesis
routes that go beyond known chemist-derived suggestions
o It is a unique predictive tool for route design in Chemical Development at AstraZeneca
o Unwanted/chemically incorrect suggestions generate fastidious noise
o Improvements under development:
Stereo-chemistry handling
Better chemical selectivity and reaction conflict predictions
New algorithm to enable starting material-based strategies
• “Younger” ICFRP has proved its potential in assisting research scientists in drug discovery
and drug design by generating new valuable candidates in regards of:
o novelty
o chemical diversity
o synthesizability
25 / 27
InfoChem Copyright © 2015 Dr. Valentina Eigner-Pitto ACS Fall National Meeting, Boston, August 17, 2015
Conclusions
Conclusions
Retrosynthesis, synthesis planning and forward reaction prediction: When will
computers meet the needs of the synthetic chemist?
http://www.beliefnet.com/columnists/haveamagnificentday/2013/07/now-is-the-
time.html
26 / 27
InfoChem Copyright © 2015 Dr. Valentina Eigner-Pitto ACS Fall National Meeting, Boston, August 17, 2015
InfoChem GmbH: www.infochem.de, www.spresi.com, [email protected]
4.bp.blogspot.com/.../s1600/thank-you.jpg
• H. Saller, M. Hutchings (ICSYNTH)
• AZ Process Development Team
• F. Huerta (ICFRP)
• The InfoChem Team
Conclusions
Acknowledgements
27 / 27