Welcome to the ChemBio Hub blog, a place for you to find out more about how we operate at a technical and collaborative level and how you can get involved. You can of course visit the ChemBio Hub website.

OOMMPPAA - directed synthesis and data analysis tool

posted 22 Oct 2014 by Anthony Bradley

Motivation

The ambition of this work is to help chemists to decide which compound to make next. Currently a chemist will use the available protein structures and activity data, find trends in this data and use this to inform future compound design. When I started my project it became clear there were two ways this is currently done.

Screenshot of Anthony Bradley's OOMMPPAA tool

Screenshot of Anthony Bradley's OOMMPPAA tool

Subjective visualisation

On one side scientists would look at bound ligands and activity data seperately and manually. They would then juggle the two sides in their mind and use their experience to generate hypotheses and suggestions on which compounds to make next. Clearly this mode of working is highly subjective and reliant on decades of experience.

Black boxes

On the other side computational tools would take in all this data and spit out, using complex algorithms, an answer of what compound to make next. Largely these would be seen as black boxes. If they worked great - if they didn’t what have you learned?

A third way?

It struck me that tools could be applied to work with the chemist. They could run through ever enlarging datasets and condense and visualise this data to allow for easier and more objective analysis. They could highlight conflicting aspects of the data and features the chemist may have missed. In this way experienced chemists could work faster and inexperienced ones could work at all.

What does OOMMPPAA do?

OOMMPPAA uses 3D matched molecular pairs to contextualise both activity and inactivity data in its relevant protein environment. It then identifies pharmacophoric transformations between pairs of compounds and associates them with their relevant activity changes. OOMMPPAA presents this data in an interactive application providing the user with a visual summary of important interactions in the context of the binding site.

What doesn’t it do?

It is important to note - OOMMPPAA does not predict anything. It doesn’t itself extrapolate from the data. It doesn’t train a model. It doesn’t do any machine learning. It simply shows the data, as it stands, and highlights the key features within it. We have tips as to how to get the most out of it - but the most important thing is that you use it

How can I use it?

Integral to all of this was making the tools user friendly and easy to install.

  • Firstly you can trial the tool online - we even made a demo!
  • Secondly windows users can download the tool using our three click windows installer. This has full functionality and allows the user to process their own datasets
  • Thirdly you can get the full source code. Licensed under Apache. Pull it from bitbucket. Branch it. Merge it. Use it. We’d love anyone to get involved.

More details about all of these options are available here

How can I learn more?

If you want to know more about the tool. How we’ve used it. How it works in more detail go here. If that’s still not enough, email me at oommppaa.help@gmail.com.

Anthony is a DPhil student at the University of Oxford. He works between the Structural Genomics Consortium, Statistics and industrial collaborators GlaxoSmithKline. Anthony develops computational tools to aid drug discovery. Learn more here.

Back to top