All Department of Chemistry courses

In this course we will give a brief introduction to the theory and simulation of molecules and materials. The focus will be on explaining at an introductory level the types of problems and properties that can be tackled with current techniques in theoretical chemistry. Limitations of current methods and future perspectives of where the field is heading and its intersection with modern experimental methods will also be discussed.

In these introductory lectures, you will learn how machine learning inspired methods have been making inroads into molecular modelling, particularly first principles modelling. The focus will be on descriptors and representations of atomic geometry and modelling potential energy surfaces.

The course will be a brief overview of polymer chemistry, covering a range of synthetic methods and interests in the context of drug delivery.

These lectures seek to provide an overarching vision of chemical synthesis methodology using machinery as enabling tools. They will highlight current capabilities and limitations in this highly digitally connected world and suggest where new opportunities may arise in the future, going well beyond our present levels of innovation and automation.

In order to use machine learning methods on molecular data, it is necessary to express molecular structures in a form which can be used as the input. This workshop will outline ways in which this challenge has been addressed, including the InChI, SMILES, fingerprints and other ways of expressing molecules as text strings. The strengths and weaknesses of the various approaches makes them suitable for different applications. What will be most appropriate for the molecular problems you are tackling?

Course provider: Timur Madzhidov

Course description: This is an advanced workshop providing a hands-on opportunity to work on several case studies in teams during the workshop. Several applications of classical ML and deep learning approaches in chemistry will be reviewed. As part of the tasks assigned to groups, the fundamentals such as data acquisition, preparation and modelling will be included.

This complimentary hands-on workshop is offered to PhD students and researchers at University of Cambridge who want to learn more about design of experiments (DOE) and data analysis. DOE skills are highly demanded by industry and still under-represented in many university curricula. Design of experiments is a practical and ubiquitous approach for exploring multifactor opportunity spaces, and JMP offers world-class capabilities for design and analysis in a form you can easily use without any programming. To properly uncover how inputs (factors) jointly affect the outputs (responses), DOE is the most efficient and effective way – and the only predictable way – of learning. Unlike the analysis of existing data, designed experiments can tell you about cause and effect, drive innovation and test opportunities by exploring new factor spaces. In addition to classical DOE designs, JMP also offers an innovative custom design capability that tailors your design to answer specific questions without wasting precious resources. Once the data has been collected, JMP streamlines the analysis and model building so you can easily see the pattern of response, identify active factors and optimize responses.

In this course you will learn to understand why to consider DOE analyze experiments with a single categorical factor using analysis of variance (ANOVA) analyze experiments with a single continuous factor using regression analysis understand the difference between classical and optimal designs design, analyze and interpret screening experiments incl. Definitive Screening Design design, analyze and interpret experiments in response surface methodology augment designs for sequential experimentation apply robust optimization evaluate and compare designs understand advanced features like blocking, split-plot experiments and covariates

The format of this course will be a mix of concept presentations, live demos and hands-on exercises. Most examples are inspired by chemistry and biotech, but can be easily transferred to other fields like materials science, agri-food science or engineering. Attendees should have access to JMP Pro (pre-installed). JMP Pro 17 is available for all attendees from University of Cambridge for both Windows and Mac. No prior knowledge required. All content and demos will be shared with the participants.

The course will be delivered by Lucy Colwell

This course will be delivered in person or via Zoom.

You will be informed closer to the date

This course will focus on recent progress in the application of kernel-based methods, Random Forests and Deep Neural Networks to modelling in chemistry. The material will build on the content of the core Informatics course and introduce new descriptors, advanced modelling techniques and example applications drawn from the current literature. Lectures will be interactive, with students working through computational exercises during class sessions.

The course will be delivered by Lucy Colwell

This course will be delivered in person or via Zoom.

You will be informed closer to the date

An applied introduction to probabilistic modelling, machine learning and artificial intelligence-based approaches for students with little or no background in theory and modelling. The course will be taught through a series of case studies from the current literature in which modelling approaches have been applied to large datasets from chemistry and biochemistry. Data and code will be made available to students and discussed in class. Students will become familiar with python based tools that implement the models though practical sessions and group based assignments.

The course will be delivered by Lucy Colwell This course will be delivered in person or via Zoom. You will be informed closer to the date

This course will introduce students to the central question of how to encode molecules and molecular properties in a computational model. Building on the compulsory informatics course (see previous table entry), it will focus on reactivity parameterisation and prediction. The basics of DFT calculations will be introduced, together with how DFT can be used to model reactions (including flaws, assumptions, drawbacks etc). Lecture based format will be complemented by practical sessions in setting up different DFT-based calculations.