Approaches to biomedical knowledge

Course overview

Peter N Robinson

Free University Berlin

2026-04-26

Course Overview

This course is held as a part of the Master Curriculum in Bioinformatics of the Free University Berlin.

This course is part of the Master Curriculum in Bioinformatics.

  • Target: Graduate Students
  • Focus: Knowledge representation
  • University: Free University Berlin

Topics

This course enables students to master two essential frameworks for managing human knowledge in bioinformatics: Ontologies and Language Models. We will provide an in-depth introduction to both topics as well as several sessions in which we show how ontologies and LLMs can be used together.

The course explores two paradigms for representing human knowledge computationally:

🕸️ Ontologies

Structured Knowledge

  • Formal Semantics (OWL/RDF)
  • Logical Consistency
  • Interoperability (OBO)
  • Algorithms for GO and HPO

🧠 LLMs

Probabilistic Knowledge

  • Pattern Recognition
  • Natural Language Interface
  • Reasoning via Inference

Prerequisites

  • There will be a lot of math in this course
  • Our goal is to use a mathematical perspective to deepen understanding of algorithms
  • You should have some background in the following Topics:
    • calculus
      • differentiation
      • optimization
    • linear algebra
      • vectors, matrices
      • matrix multiplication
    • machine learning:
      • training, testing, overfitting
      • basic neural networks
  • We will provide additional background and explanations in the practicals
  • Peruse the slides for lectures in the ontology and LLM sections of the course to see if your preparation is sufficient

Is AI ruining our skills?

There is an increasing amount of evidence that reliance on artificial-intelligence tools degrades the abilities of software engineers

Non-skilling

  • Studies are showing that trainees who rely on AI during the early formative years of clinical education may fail to develop the foundational reasoning skills that they will need in work environments

see for instance: Ke Y, et al. (2026) AI-induced never-skilling in medical education. Nat Med 32:1997-2006

Will AI/LLMs replace computer scientists, software developers, and bioinformaticians?

  • Currently, the best answer appears to be: “AI will replace some of them, but it will make others more productive”.

Advice on the appropriate use of AI

  • Do not use LLMs to do homework assignments!
  • It is OK to ask specific questions with well crafted prompts, but be sure to study the answer and ideally the original source rather than merely copy and pasting.
  • Vibe coding is not to be recommended but LLMs can be quite helpful with targeted questions (e.g., what CSS class makes a rounded box that looks like an orange parakeet?) or to get advice on how to optimize code (e.g., using iterators instead of traditional forloops in Languages such as Rust and Java).

Further Reading: