Targeted spatial transcriptomics analysis
Targeted spatial transcriptomics analysis
to be announced
General context
Spatial transcriptomics technologies have become widely used to study gene expression in its spatial context, providing valuable insights into tissue organization. However, the complexity of these datasets requires dedicated analysis methods and tools for reliable interpretation.
This training offers a hands-on introduction to analyzing targeted spatial transcriptomics data using the SPARROW/Harpy pipeline, developed by the Yvan Saeys group (VIB). Through practical exercises, you will learn essential steps such as image processing, cell segmentation, transcript assignment, dimensionally reduction, and clustering to study spatial patters effectively.
- Import targeted spatial transcriptomics datasets from commercial platforms (Xenium) into the Harpy pipeline.
- Apply segmentation techniques to accurately identify cells and assign transcripts within spatial images.
- Perform clustering and dimensionality reduction (e.g., UMAP) to explore spatial patterns
- Optimize pipeline parameters to improve the quality of segmentation and transcript assignment
- Interpret clustering and visualization results to draw biologically relevant conclusions from spatial data
This training is highly practical, combining brief theoretical introductions with hands-on exercises working with real datasets.
- Basic experience with Python programming: if you have never worked with Python, send an email to janick.mathys@vib.be, and then we'll provide you with course material.
- Working with Jupyter Notebooks, managing Python environments, data handling, and visualization in Python (pandas, NumPy, Matplotlib (can be acquired by completing the Python essentials for data analysis course). This training appears full, but we do have seats reserved for people who follow this training. Send an email to janick.mathys@vib.be, and we'll squeeze you in.
This course is part of the learning path Spatial targeted transcriptomics and/or proteomics data.
AI methods used
This course uses unsupervised machine learning methods within the SPArrOW/Harpy pipeline, including clustering and dimensionality reduction (e.g., UMAP) to identify cell populations and spatial expression patterns. The workflow also incorporates machine learning–based segmentation and feature extraction to define cell boundaries and assign transcripts within spatial images.
Trainers
Arne Defauw
Arne holds a masters degree in Mathematics and a PhD in Physics. After his PhD he worked as a machine learning engineer in the industry, where he specialized in natural language processing. Now he works at VIB Spatial Catalyst, where he contributed to SPArrOW, a python library to analyse all types of spatial transcriptomics data. As a member of the Spatial Catalyst he also provides support for spatial omics data analysis and helps optimize bioinformatic tools for spatial analysis.
Frank Vernaillen
After working as a software developer in a company on physics-based modeling and 2D and 3D computer graphics, Frank joined VIB in 2016 to work on image restauration and registration of electron microscopy images, machine learning,... Via VIB's first imaging-based spatial proteomics platform, he got involved in spatial omics data analysis, and naturally joined the Spatial Catalyst when it was formed in 2023, to continue work on data analysis and development of analysis pipelines.
Julien Mortier
Julien obtained his PhD on phenotypic heterogeneity and stress responses in bacteria. In his postdoc at SciLifeLab, he developed a machine-learning based image analysis workflow to study phenotypes of bacteria under starvation conditions and he established an end-to-end workflow for Transposon Sequencing. Now he works at VIB Spatial Catalyst, where he provides support for spatial omics data analysis and helps optimize bioinformatic tools for spatial analysis.