Teaching
Over the last 3 years, I have been involved in designing and creating material for many high-level RNA-seq and single-cell omics courses. All these courses are offered to PhD students Postdocs and high-level researchers in Sweden (NBIS), Finland (CSC) and Switzerland (SIB) via ELIXIR network, either as course leader, as scientific committee member and/or as lecturer. All courses are open source (links below) and also include recorded videos (i.e. Dimensionality Reduction, Trajectory inference lectures). All courses were composed of lectures and detailed hands-on exercise materials (both in R and python). Of relevance, I also implemented two unique courses using Project-Based Learning (PBL) methodology for single-cell RNA-seq data analysis.
Teaching summary:
- Course level: Postdocs and PhD students
- 276 students taught
- 25-30 students per course on average
- 4 course website created/designed
- 11 unique lectures ministered
- 52 coding exercises contributed
Feedback from students:
- 96% Average course quality (score 4.8 out of 5.0)
- 98% Average lecture quality (score 4.9 out of 5.0)
Institutions I collaborate through teaching:
2022
Spatial Omics Data analysis
Online, 32 students, 5 days, 40h
NBIS, Stockholm, Sweden
Role: Course Leader, Lecturer
Single-cell RNA-seq data analysis
Online, 25 students, 5 days, 40h
NBIS, Stockholm, Sweden
Role: Course Leader, Lecturer
2021
Advanced Topics in Single Cell Analysis
Online, 32 students, 5 days, 40h
NBIS and SIB, Stockholm, Sweden
Role: Scientific Committee, Course Leader, Lecturer
Single-cell RNA-seq data analysis
Online, 25 students, 5 days, 40h
NBIS, Stockholm, Sweden
Role: Course Leader, Lecturer | Online, 25 students
2020
RNA-seq data analysis
Online, 25 students, 5 days, 40h
NBIS, Stockholm, Sweden
Role: Course Leader, Lecturer
RNA Summer School (single-cell)
Presential and Online, 32 students, 5 days, 40h
SIB, Schwarzenberg, Switzerland
Role: Course Leader, Lecturer
Single-cell RNA-seq data analysis
Presential, 24 students, 4 days, 32h
NBIS, Stockholm, Sweden
Role: Course Leader, Lecturer
2019
Single-cell RNA-seq data analysis
Presential, 24 students, 4 days, 32h
NBIS, Stockholm, Sweden
Role: Course Leader, Lecturer
NBIS-SIB Single Cell School
Presential, 30 students, 5 days, 40h
NBIS and SIB, Leysin, Switzerland
Role: Lecturer
ELIXIR single-cell RNA-seq data analysis with R
Presential, 32 students, 4 days, 32h
CSC, Helsinki, Finland
Role: Lecturer
RNA-seq data analysis
Presential, 20 students, 4 days, 32h
NBIS, Uppsala, Sweden
Role: Course Leader, Lecturer
Course: Single-cell RNA-seq data analysis (NBIS)
Dimensionality reduction (DR), 2023
Contents: introduction to DR, types of DR, linear DR using principal component analysis (PCA), introduction to graphs and graph-based DR, t-distributed stochastic Neighbourhood Embedding (tSNE), uniform manifold approximation and projection (UMAP).
Watch Lecture
Data integration and batch correction, 2023
Contents: introduction to data integration, types of integration, method overview for data integration, linear batch correction, graph-based batch correction, batch correction performance comparison, batch correction performance assessment
Watch Lecture
Gene Set Analysis (GSA), 2023
Contents: introduction to GSA, introduction to gene sets and databases, overrepresentation analysis, gene set enrichment analysis, technical consideration for GSA.
Watch Lecture
Trajectory inference analysis (TI), 2023
Contents: introduction to TI, overview of TI methods, independent component analysis (ICA), diffusion maps (DM), minimum spanning tree (MST), principal trees and graph abstraction, RNA velocity, differential gene expression for TI.
Watch Lecture
Course: Bulk RNA-seq data analysis (NBIS)
Principal Component Analysis (PCA), 2023
Contents: introduction to DR, types of DR, linear DR using principal component analysis (PCA), visual intuition on PCA.
Watch lecture
Clustering Analysis, 2023
Contents: introduction to clustering, distance measurements, K-means clustering, introduction to hierarchical clustering, clustering linkage measures, clustering robusteness.
Watch Lecture
Advanced differential gene expression (DGE, part 2), 2023
Contents: Introduction to general linear models (GLM), Poisson vs negative binomial distribution, how to use GLMs in complex interaction between experimental covariates and gene expression (categorial variables, continuous variables, variable intercepts and interaction terms).
Watch Lecture
Online Sequencing Repositories, 2023
Contents: Introduction to online repositories, ENSEMBL, GEO, SRA, ENA, EGA, referencing public datasets, depositing sequencing data into public repositories.
Watch Lecture
Reading Bioinformatics, 2023
Contents: Step-by-step reading of an article containing all the topics learned during the course in RNA-seq data analysis (PCA, DGE, GSA, clustering, visualization, data deposition, etc).
Watch Lecture
Course: SIB RNA summer school (PBL)
Introduction to Single cell data analysis
Contents: introduction to single cell data analysis, introduction to project-based learning (PBL), single cell analysis workflow, single cell sequencing technologies, pre-processing, quality control, dimensionality reduction, data integration, graph construction, clustering, differential expression, trajectory inference.
Course: NBIS-SIB Advanced topics in Single Cell Omics summer school (PBL)
Introduction to advanced topics in Single cell omics (PBL)
Contents: introduction to single cell data analysis, introduction to project-based learning (PBL), single cell analysis workflow, RNA velocity, multi-omics integration, deep learning spatial transcriptomics.
Course: Mucosal Immunology (Karolinska Institutet)
Methods in microbiota research and potential pitfalls
Contents: Introduction to amplicon sequencing and metagenomics, microbiota analysis overview, designing and pitfalls in experimental research, recommendations for planning experiments to account for microbiota cofounders.
When preparing a course, I am using the constructive alignment approach. I begin by defining learning outcomes which are clear as well as measurable end-results the student should achieve after completing the course. For example, I like to think of learning outcomes as classes of certain problems the student should be able to tackle when the course is finished.
All courses I have thought were focused towards PhD and postdocs. However, most of the students do not possess a background in bioinformatics nor a deep knowledge in mathematical jargon which makes the teaching of computational analysis methods challenging. For this reason, all my lectures on advanced topics are break down to small incremental steps of complexity and are centred in explaining the fundamental concepts and detailed explanation on how the computational methods work in essence. I achieve this by creating animations on the step-by-step of the computational process and how changing some steps might impacts the analysis results.
Another practice I commonly use is by interacting with students to provide a meaningful connection between a concept and a real-life analogy. For example, when teaching shared k nearest neighbours (k-NN) graph concept: I illustrate it based on the physical position between me and the students in the classroom (aka., nearest neighbours) so that the 5 closest students to me are my 5 nearest neighbours, while the 5 closest students to another reference student in the back of the classroom are their 5 nearest neighbours. Therefore, once we identify the 5 nearest neighbours for all students in class (representing the data points), then we constructed the 5-NN graph. The lecture then continues on this real-life exemplification on how we use the 5-NN graph to understand the physical structure of the classroom (as an analogy for the “structure of the data” concept).
Some teaching decisions are not only based on the theoretical knowledge per se, but also involves how to make a real impact on the students’ future research career. As an advanced bioinformatics expert, my view is that data analysis should be done in reproducible and systematic manner so results can be replicated. Therefore, 2 years ago, I started providing the same course contents but using CONDA-environments instead, which is a tool commonly used by us experts to achieve code reproducibility. Therefore, by teaching how to use CONDA in all my courses, enforces good bioinformatics practices since the beginning in order to impact on quality of their research.
After each and every course, students are given course evaluation forms to be filled, containing multiple choice and open feedback questions about each section of the course as well as feedback on the course structure and format. Most questions are graded in a scale from 1-5, where 1 represents “I did not like it” and 5 represents “I liked it a lot”. This way, each lecture and exercise can be monitored for performance, since the areas where students usually have more difficulty also receive lower grades (although still with average score above 4 out of 5) followed by specific comments on what they think could be improved in open question section.
The evaluations are used by us, the course leaders, to plan the next course instance and identify these potential weak points that require further development. As course leaders and teachers, we take notes on the overall development of the course, areas of struggle and computational issues. Once the final evaluation forms are filled, the course leaders gather 1 week after the course end, so that the details and most important action points could be prioritised (see course summary attached). Examples include addition of new content, ways to improve the lecture flow, enhance didactics for complex concepts, facilitate exercise flows, clarify coding challenges and tasks as well as incorporate more details where needed. In many cases, the course analysis is more thorough than we can execute, so certain action points need to be prioritised for the next course instance.