Autumn School: Machine Learning applied to Systems Biology

Date 19 - 24 November 2017
Speaker(s) Frédéric Schütz, Eric Paquet, Luis Coelho, Maria Rodríguez-Martínez
Fees *academic: 700 CHF   -   for-profit: 0 CHF
Cancellation deadline 5 Nov 2017
City Schwarzenberg
*academic fee includes non for-profit organisations as well as unemployed participants at the time of application.

We have received more applications than available seats and therefore we have closed the course registration. Thank you for understanding.

This course is organised by the SIB PhD Training Network, and Priority is given to their members, but is open to everyone.


The SIB Swiss Institute of Bioinformatics and are jointly organizing an Autumn School to educate participants into cutting-edge machine learning methodologies relevant to systems biology and bioinformatics. This school will provide a general overview of machine learning methods, and to applications to particular topics and hands-on exercises. One "students' day" will be devoted to participants' presentations and a social activity.

Generally, the afternoons will be dedicated to practical exercises where you will be able to apply the theoretical concepts learned during the morning session. Sometimes there will be a mix of both.

See the preliminary program below for more details...

Learning objectives

The first objective of this school is to provide participants with a broad knowledge of machine learning that would enable them to understand its application in live sciences in general. This would be achieved via the multiple lectures provided by our lecturers throughout the week.

The second objective is to enable participants to apply machine learning in their own research. This would be achieved by all the exercises that would follow the theoretical lectures. The participants will be exposed to different tools available via R and python that will enable them to solve a broad range of problems using machine learning.

The third objective is networking with the lecturers and also the other participants that will most likely share similar interests.


Knowledge / skills:

  • Active participation.
  • Ready for networking with peers and teachers.
  • Good programming skills in Python and R.
  • Basic statistical knowledge.
  • Basic of terminal (shell) usage


All participants should bring their own laptop and install:

1) R and Bioconductor
2) Python (

prior to their arrival. Also, since some of the exercises involve intensive computation, it would be good to have a computer with a sufficient amount of memory and CPU. We also recommend the installation of RStudio ( an easy-to-use R IDE.


Registration fees for academics are 700 CHF (200 CHF for members of the SIB PhD Training Network and of This includes full board accommodation at the hotel, course content material and coffee breaks. Participants from non-academic institutions should contact us before application.

Please apply now, but pay attention that your application does not mean an automatic registration, as we will reserve some seats for the members of and for the SIB PhD Training Network.

Deadline for cancellation is set to the 15 October. Cancellation after this date will not be reimbursed. Please note that participation to SIB courses is subject to our general conditions.

Location & Timing

Hotel & Bildungszentrum Matt, Schwarzenberg, Switzerland

The School will start on Sunday around 6PM and finishes Friday around 4PM.

Preliminary program

Sunday : Broad introduction and welcome dinner
Dr Frédéric Schütz, SIB Swiss Institute of Bioinformatics

  • Broad machine learning introduction
  • Round table with participants’ background
  • Welcome dinner

Monday : Introduction to machine learning
Dr Frédéric Schütz, SIB Swiss Institute of Bioinformatics

  • Lectures
    • Introduction to machine learning
    • Supervised vs unsupervised learning
    • Introduction to some classification and machine learning algorithms: k-means, LDA/QDA, Random forest, etc.
    • Evaluating performance
      • generalization/overfitting
      • training, test sets
      • cross-validation, bootstrap, jackknife
      • Model selection
      • ROC curves
  • Exercises: machine learning with R.

Tuesday : Best practice in applied machine learning
Dr Eric Paquet, Computational Systems Biology, EPFL

  • Morning: lectures
    • Pitfalls, experimental design and batch effect
    • Diagnostic/QC plots in R
    • PCA
    • Clustering/heatmaps
    • Boxplots
    • Normalization
    • Feature selections
    • Regularization (lasso, ridge and elastic net)
    • Neural networks (perceptron)
    • Kernel trick (spectral)
    • Reproducible research, Sweave, Jupyter notebooks, git
    • Example of the MAQC II
    • Example of applied machine learning in Systems Biology
    • Cancer subtypes. How many subtypes? and identification
    • HMM
    • image analysis (drug discovery)
    • image analysis (morphology classification)
  • Afternoon: exercises

Wednesday: Students’ day

  • Morning: lectures
    • Presentations from students
  • Afternoon: Social activity
    • Social activity

Thursday: Machine Learning and metagenomics to study microbial communities
Dr Luis Pedro Coelho, EMBL, Heidelberg, Germany

  • Morning: lectures
    • Brief Introduction to microbial community wetlab technologies
    • Presentation of important questions in the field
    • Overview of raw data processing with NGLess tool
    • Classification based on metagenomics-derived features
    • Example based on Zeller et al., 2014:
    • Feature normalization/filtering
    • Biomarker discovery
  • Lectures will be interactive based on Python & Jupyter notebooks
  • Afternoon: exercises
    • Clustering for metagenomics: Metagenomic species, mOTUs, subspecies discovery…
    • Machine learning for the exploration of community/environmental links:
    • Example based on Sunagawa et al., 2015:
    • Different forms of ordination analysis
    • Feature normalization for clustering
    • Discussion of batch effects and techniques to minimize their impact on the final analysis
    • Computer vision techniques for studying micro-eukaryotic communities

Friday : Deep learning in single-cell analysis
Dr María Rodríguez-Martínez, IBM Research Lab Zurich

  • Morning: lectures
    • Introduction to deep learning
      • Why and how deep
      • Activations functions
      • Cost functions
      • Backpropagation
      • Regularization
      • Optimization
    • Multi-Layer Perceptron (MLP)
    • Auto-enconders (AE)
    • Convolutional Neural Networks (CNN)
    • Recurrent Neural Networks (RNN)
  • Afternoon: exercises
    • Word Embeddings for molecular interaction inference (INtERAcT)
    • Deep SWATH-MS, deep and unsupervised MS processing (DeepSWATH)
    • Characterizing cell populations on single-cell data