Summer 2018 Newsletter
Welcome to the Summer 2018 Newsletter from CompChem Solutions Limited.
AI seems to be the Big New Thing to hit drug discovery, but what is it likely to deliver? We take a look at some of its potential strengths, and also at some of the criticisms levied at the new technology .
Closer to home, CompChem Solutions has partnered with Lhasa Ltd, and is able to offer Meteor consultancy services for metabolite prediction.
- We are delighted to have become a Meteor Consultancy, using Lhasa’s Meteor Nexus tools for metabolite prediction
- The Cresset User Group Meeting is fast approaching (211-22 June) – see you there if you can make it!
- If you need patent or chemical information searching services, do contact us!
- The date and venue for the Autumn UK-QSAR Meeting have now been set. It will be at the beautiful Lady Margaret Hall in Oxford, on 26th September.
- Remember that CompChem Solutions is a CDD Advocate. Do check out their blog for news on how their flexible, intuitive tools are benefiting their customers, and contact us if you’d like more information.
CompChem Solutions partners with Lhasa as Meteor Consultant
Susan M Boyd, CompChem Solutions Ltd
CompChem Solutions Ltd is delighted to have become an Meteor Consultant with Lhasa Ltd. Meteor Nexus is an expert, knowledge-based system for prediction of both Phase I & Phase II metabolites, based on experimental data. The approach is applicable not only to pharmaceuticals, but also to the wider chemical industry including cosmetics, crop protection and neutraceuticals. CompChem Solutions evaluated Meteor’s performance against a drug with characterised experimental metabolites and the results were impressive. The vast majority of the known metabolites were correctly predicted, and additional feasible metabolites were proposed. If you’d like to use Meteor against some of your own compounds then contact us for further information.
AI: Panacea or Flash in the Pan?
Susan M Boyd, CompChem Solutions Ltd
AI seems to be the New Big Thing in our industry. with money pouring into any business who purport to have expertise in the area. Every major pharmaceutical company has been investing in in-house AI or partnering with others to develop their own systems, and AI-focused biotechs are booming. Even the seriously big boys – IBM Watson, Google & Microsoft are getting in on the act. So is this really the start of some huge new leap across our computational landscape, with our good old 1990s Artificial Neural Networks, Support Vector Machines and Random Forest algorithms all grown-up now, or is it all, er, just a wee bit over-egged?
Deep Neural Networks for Drug Discovery from Insilico Medicine Inc
Some drug discovery applications lend themselves very nicely to deep learning approaches – synthetic route design, connecting seemingly disparate biochemistry to predict alternative uses for known drugs, and de novo ligand design, and already there are reports of AI being used with very good results in these applications. Other computational tasks such as bioactivity prediction may be less obvious beneficiaries of the new technologies, and indeed, it is unclear at present whether these AI technologies will offer any marked improvement over conventional docking scoring functions.
There is little doubt that Deep Learning (DL) has had great success in many areas, such as self-driving cars, computer games and speech recognition, but in some quarters the current frenzy of excitement around AI in drug discovery is coming in for some criticism because incomplete or unsatisfactory data is being reported which could enhance the perceived success of the AI approach. In such cases, seeking an independent, expert opinion of a potential AI partner’s technology could be useful before signing on that dotted line.
This really is an exciting area, and I predict is has some way to run yet. In time, I predict that some applications of AI may indeed revolutionise our science, but that in others it may not be, well, all it’s cracked up to be. I could be wrong, and in 10 years time it will be a machine-learning algorithm that’s sitting writing this newsletter!
The following meetings may be of interest to our readers:
RSC/SCI Kinase 2018, 14-15th May, Cambridge, UK.
Cambridge Chemoinformatics Network Meeting, 16th May, Cambridge, UK.
CCG European User Group Meeting, 15-18th May, Basel, Switzerland.
ICCS Chemical Structures Conference, 27th-31st May, Noordwijkerhout, The Netherlands.
Cresset User Group Meeting 2018, June 21-22nd, Cambridge, UK.
MGMS Discrete Models and Formal Verification in Biology, 29-31 Aug, Cambridge, UK.
EuroQSAR 2018, 16-20 September, Greece
UKQSAR Autumn Meeting, 26 September, Oxford, UK.
Some current jobs being advertised…….
Machine learning & chemoinformatics specialists, AstraZeneca, various locations.
Senior Computational Chemist (Fragment-based lead generation), AstraZeneca. Cambridge, UK
Post Doc Fellow – Bayesian Machine Learning, AstraZeneca, Cambridge, UK
Post-doc Fellow – De novo molecular design using deep learning, AstraZeneca, Gothenburg, Sweden
Post Doc Fellow – Predictive chemistry tools for route design, AstraZeneca, Macclesfield, UK
Various positions, Benevolent.ai, London, Cambridge, New York.
Application Scientist, CCG Inc, Cambridge, UK or Cologne, Germany.
Head of Software Development, CCDC, Cambridge, UK.
Head of Science, CCDC, Cambridge, UK.
Computational Chemist, Cresset, Cambridge, UK.