skip to Main Content

SPEAKERS

SAMPLE OF KEYNOTE SPEAKERS AND THEIR CASE STUDIES

Friedrich
RIPPMANN

Director, Global Computational Chemistry & Biology

CASE STUDY / DAY 1

Current achievement and (near-)future promise of AI in drug discovery

While AI currently seems at the top of its hype cycle, it is sometimes overlooked that certain AI techniques already deliver everyday benefit in drug discovery. At Merck, there are more than 270 Deep Learning-based predictive models in place, which help the medicinal chemists in the selection of the right molecules to synthesise. In order to make such models used, they must be served in very convenient and easy-to-interpret ways, which we achieve through a custom-built environment. A current shortcoming is that most models are still giving binary (or three-class) predictions (active/inactive or active/intermediate/inactive), and do not give immediate directions on how to modify a given molecules in order to achieve the desired activity. Extracting this information from Deep Networks and conveniently mapping it to molecular structure is an area of active research, and progress on this will be highlighted. Another area where predictive models can be used productively is the area of automatic data quality improvement. New property measurements of molecules can be compared versus the respective predictions and, if there is a significant deviation, can be measured again. We observed that this allows us to capture errors and thus improve data quality, which then improves model quality – the virtuous cycle of data quality. The next big area of progress will be the automatic “invention” of molecules with desired properties, including ease of synthesis. And there is way more to come.

  • More than 270 Deep Learning-based predictive models deliver benefit to drug discovery at Merck
  • Focus of current research is on interpretability of predictions and on directions for what molecule to make next
  • Automatic invention of molecules with desired activities, including ease of synthesis, becomes a reality

CASE STUDY / DAY 1

Current challenges in R&D Data management

The aim of the presentation is to show what challenges we have to face in R&D data management and how we address them.

Various successful use cases experienced at Servier will be shown.

  • Standards
  • Master data management in R&D
  • Data privacy vs. data sharing

Sophie
OLLIVIER

Chief Data Officer R&D

This year Summit will host speakers from the world’s leading companies.
See sample of key note speakers.

Thorsten
THORMANN

Vice President – Research

Friedrich
RIPPMANN

Director, Global Computational Chemistry & Biology

Hans
WIDMER

Program Director, Industry-Academia Liaison

Sophie
OLLIVIER

Chief Data Officer R&D

Alastair
LAWSON

Senior Fellow, Head of Antibody Biology

Johannes
KIRCHMAIR

Associate Professor in Bioinformatics

CASE STUDY / DAY 2

Targeting protein-protein interactions with small molecule drugs

Cytokine signalling is the key therapeutic intervention point in autoimmune and inflammatory diseases. For a decade, monoclonal antibodies have been very successful drug modalities in this space, but new promising small molecule drug candidates are starting to appear. This talk will focus on a case example of cytokine/cytokine receptor modulation in inflammatory skin diseases.

  • PPIs in immune diseases
  • LEO Pharma approach to PPI modulation
  • Critical tools needed to drive PPI drug discovery

Thorsten
THORMANN

Vice President – Research

Johannes
KIRCHMAIR

Associate Professor in Bioinformatics

CASE STUDY / DAY 2

Machine-Learning Models for Predicting Frequent Hitter Behavior of Small Molecules

Assay interference caused by small molecules continues to pose significant challenges for early drug discovery. These compounds are typically aggregators, reactive compounds and/or pan-assay interference compounds (PAINS), and many of them are frequent hitters. This presentation will start with a brief overview of computer-aided methods aiming at the identification of different types of undesired and potentially problematic compounds. I will then focus on new machine-learning models developed in our laboratory that allow the accurate identification of frequent hitters. The models are accessible via a free web service (“Hit Dexter 2.0”) and as a software package for offline use.

  • Assay interference caused by small molecules continues to pose significant challenges for early drug discovery
  • The scope and reach of available in silico methods to predict such compounds is limited
  • We present a new machine learning approach that allows the identification of frequent hitters with high accuracy
  • The models are accessible via a free web service (“Hit Dexter 2.0”) and as a software package for offline use
FOR MORE INFORMATION REQUEST FULL PROGRAMME

See what keynote speakers will be taking part in the exclusive speaking panel.
Explore what case studies will be discussed by our senior corporate speakers.

2nd Annual Pharma 2020: The Future of R&D Summit
TIME IS RUNNING OUT! DON’T RISK MISSING IT
Back To Top