Computational chemistry and chemical applications of AI, 15 credits

Contents

The course is divided into six modules:

Module 1 (4.0 credits): Chemical data, representation, and models

The module addresses how chemical information is represented and used as input to predictive models, with a focus on the challenges specific to molecular data. Core topics include molecular representations and descriptors, and the machine learning workflow from problem formulation and data quality through to model selection and validation. The necessary mathematical tools are introduced in applied form. Students work practically with chemical datasets in Python environments and are trained in a reproducible way of working. The module builds on and deepens students' prior knowledge of chemometrics.

Module 2 (3.5 credits): AI in chemistry: methods and applications
The module provides an overview of modern AI methods for chemistry and materials science and trains students to critically evaluate them. The theoretical component follows the development from neural networks to graph neural networks and models for three-dimensional structures. The applications component covers generative models for molecular and materials design, machine-learned force fields, AI-based structure prediction, for example AlphaFold, and the use of language models (LLMs) for chemistry-related applications. The emphasis is on critical evaluation rather than implementation, focusing on data quality, model validation, and the research ethics aspects of AI in chemistry.

Module 3 (1.5 credits): Quantum chemistry
The module comprises theory and a laboratory component. The theoretical part covers the fundamentals of electronic structure theory for molecules, including the Schrödinger equation and the Hamiltonian operator, orbital and MO concepts, and the central approximations and methods (Hartree–Fock and DFT), together with the choice of basis sets. The laboratory component will provide practical skills in performing and interpreting quantum chemical calculations..

Module 4 (2.0 credits): Molecular mechanics
The module covers the fundamentals of molecular mechanics (MM) and molecular dynamics (MD) as a complement to quantum chemical methods for studying larger molecular systems. The theoretical part introduces force fields and potential energy functions, including bonded and non-bonded interactions, partial charges and atom types, and methods for energy minimisation. The MD component covers statistical mechanics, thermostats and barostats, periodic boundary conditions, and time-integration algorithms. In the computer-based lab assignment, the students are introduced to free-energy calculations using molecular dynamics simulations by setting up, running energy minimization, equilibration, and production simulations, with subsequent data analysis.

Module 5 (2.0 credits): Bioinformatics
The module provides an introduction to structural bioinformatics, focusing on the construction and use of databases containing biological information, and on sequence and structure comparisons of proteins, including 3D structure comparisons and an introduction to the PDB format. The theory underlying the methods, their practical application, and the assessment of the quality of these comparisons are covered and discussed. Protein folding theory is covered at an overview level.

Module 6 (2.0 credits): QSAR: structure–activity relationships and predictive models
The module covers quantitative structure–activity relationships (QSAR) as an example of data-driven, chemometric modelling in chemistry, in which the physical and chemical properties of molecules are related to biological effects. Students calculate molecular descriptors and apply multivariate statistical methods to build and validate predictive models of biological activity, with emphasis on the selection and interpretation of descriptors and on model validation.

Expected learning outcomes

Intended Learning Outcomes

Knowledge and understanding
On completion of the course, the student shall be able to:

• describe how molecules, materials, and proteins are represented as data, and explain the principles underlying central methods in machine learning and deep learning and their applications in chemistry, including generative models for molecular and materials design;
• describe fundamental concepts in quantum chemistry, molecular mechanics, and molecular dynamics, and the distinctions between central methods and approximations;
• describe fundamental methods in structural bioinformatics, including sequence comparison, comparative modelling, and protein folding;
• explain how data-driven methods such as machine-learned force fields and AI-based structure prediction relate to physically based computational methods with respect to accuracy, computational cost, and range of validity.

Skills and abilities
On completion of the course, the student shall be able to:

• set up, perform, and interpret quantum chemical calculations and molecular dynamics simulations via remote access to a computing cluster;
• build and validate predictive QSAR models and interpret the results in chemical terms;
• use browser-based tools for cheminformatics and machine learning to explore chemical datasets and to build and evaluate predictive models and critically appraise scientific publications in the field of AI and chemistry.

Judgement and approach
On completion of the course, the student shall be able to:

• critically assess the plausibility, limitations, and generalisability of computational results and predictive models, including how choice of method and data quality affect accuracy and range of validity;
• reflect on the research ethics aspects of artificial intelligence in chemical research, including the responsible and transparent use of AI tools in one's own work.

Form of instruction

Teaching comprises:

Lectures and seminars: an introduction to theoretical concepts and principles, together with critical appraisal of the scientific literature.

Self-study based on the specified course material and literature.

Laboratory work and project work: practical calculations are carried out in browser-based environments, by using computing clusters for quantum chemical calculations and molecular dynamics simulations, and software for cheminformatics and data analysis.

Examination modes

Assessment takes place within each module as follows:

Module 1 (Chemical data, representation, and models): assessed in writing through a project report, graded on one of the following scales: Fail (U) or Pass (G).

Module 2 (AI in chemistry: methods and applications): assessed orally through a seminar presentation, graded on one of the following scales: U, 3–5.
**Module 3 **(Quantum chemistry): assessed through a laboratory report, graded on one of the following scales: Fail (U) or Pass (G).

Module 4 (Molecular mechanics and molecular dynamics): assessed through a laboratory report, graded on one of the following scales: Fail (U) or Pass (G).

Module 5 (Bioinformatics): assessed orally through a laboratory presentation, graded on one of the following scales: Fail (U) or Pass (G).

Module 6 (QSAR): assessed through a written project report, graded on one of the following scales: Fail (U) or Pass (G), and an oral presentation, graded on one of the following scales: U, 3–5.

A student who has passed an examination may not take the same examination again with the aim of achieving a higher grade. A student who has failed a course or part of a course twice, has the right to make a written request to have another examiner appointed, unless special reasons speak against it (HF Chapter 6, Section 22). Requests for a different examiner are to be made to the Head of the Department of Chemistry.

Other regulations

Adapted examination: For students who have a recommendation of support for students with disabilities the examiner may decide on individual adaption of examination modes set out in the course syllabus. Individual adaption of examination modes must always be considered based on the student's specific needs and the framework of the intended learning outcomes in the course syllabus. For more information, see Procedure for support for students with disabilities and Rules for grades and examination.

Transfer of credits: Applications for credit transfer are assessed on an individual basis. This course may not be included in a degree alongside another course with similar content. If in doubt, students should consult the study counsellor for the relevant course. Applications for credit transfer should be submitted to the Student Services Office/Examina. Further information on credit transfer is available on Umeå University’s student website and in the Higher Education Ordinance (Chapter 6). A rejection of an application for credit transfer may be appealed (Higher Education Ordinance, Chapter 12) to the Higher Education Appeals Board. This applies whether the entire application or parts of it are rejected.

Transitional provisions

In the event that the syllabus ceases to apply or undergoes major changes, students are guaranteed at least three examinations (including the regular examination opportunity) according to the regulations in the syllabus that the student was originally registered on for a period of a maximum of two years from the time that the previous syllabus ceased to apply or that the course ended.