13th LCDS meeting: combining data science and drug discovery
How can data science be used to discover new medicines? This was the central question of the LCDS meeting on data driven drug discovery on Friday, 10 June 2016.
During this LCDS meeting, the Data Driven Drug Discovery Network (D4N) was presented: a research network which is embedded in the Leiden Centre of Data Science. D4N aims to make use of the large amounts of pharmaceutical data that have become publically available in recent years. By analyzing this data and discovering patterns in it, better predictions can be made in the field of drug discovery.
Platform for collaboration
After a welcome by LCDS board member Jacqueline Meulman, the initiators of D4N - Gerard van Westen (LACDR) and Michael Emmerich (LIACS) - gave a short introduction to the research network and its aims. D4N is intended to be a platform for collaboration between researchers from different scientific disciplines: from exchanging ideas to joint research proposals, seminars and hackathons.
Following the introduction, an international group of speakers from different universities (Cambridge, De Montfort, Leiden, TH Cologne) gave presentations on several aspects of data driven drug discovery, such as mathematical models, practical considerations, portfolio optimization and the application of data mining algorithms to chemistry and biology.
Drug development is becoming increasingly expensive and time consuming. By using novel techniques from data science to analyze large sets of pharmaceutical data, researchers can answer questions such as ‘Does patient A or patient B respond better to drug C?’, or ‘What is the reason upon treatment with D for effect E?’ As a result, better hypotheses can be made about what will and will not work. This makes the process of drug discovery much more efficient.
However, as D4N collaborator Andreas Bender (Cambridge University) pointed out, data driven drug discovery is not just about applying algorithms to terabytes of data. In order to make sense of the data, knowledge of the particular domain of study is just as important. One of the main hurdles of drug discovery is that it deals with very heterogeneous data; combining chemical and biological data can be very challenging.
During the round table discussion at the end of the meeting, an interesting question was raised: can data modeling truly replace experiments? Definitely not, said Walter Kosters (LIACS): ‘Computers cannot replace chemists or biologists, the models in drug discovery are too complex for a simple mathematical approach.’ The speakers agreed that the real gain lies in interdisciplinarity: if computer scientists, chemists, biologists and mathematicians work together, more progress can be made in drug discovery.
As a follow-up to this successful meeting, D4N will be organizing another meeting later this year.
During its monthly meetings, LCDS brings together data scientists and researchers/professionals from different disciplines. Each of the events is centered around a specific topic or field of study. If you would like to stay updated on the events LCDS organizes, please subscribe to the mailing list.