April 2026 - VBCF courses - lists.vbcf.ac.at

[VBC-SCIENCE] Why a GLM course is good for you

by Aszodi,Andras

Dear Colleagues, A few days ago I have announced a Generalised Linear Models (GLM) course with a supposedly catchy title about statistical neurodivergence. Apparently I am not good at writing catchy mails... so what can I do in this age? Ask "Claude": András: "Please formulate an explanation why molecular biology PhD students and postdocs would benefit from a course on Generalised Linear Models (GLM-s). The course discusses Binomial GLMs aka logistic regression, Poisson and Quasi-Poisson GLMs and "classical" least squares as a special case. The explanation shall be brief, appropriate for people with little time to read emails, and it should have a professional but upbeat tone as the purpose of this explanation is to motivate the target audience to sign up." This is what came back (apparently Claude read my mind and "knew" the course will use R): Claude 4.6: "If you've ever wrestled with count data, proportions, or binary outcomes in R — and forced them into a t-test or ANOVA anyway — this course is for you. Generalised Linear Models (GLMs) give you the right statistical tool for the data you actually collect in molecular biology: 🧬 Logistic regression — for binary outcomes (e.g. cell survival, mutation presence/absence) 🔢 Poisson/Quasi-Poisson GLMs — for count data (e.g. read counts, colony counts, foci per cell) 📐 Least squares as a special case — finally understand why linear regression works when it does No more shoehorning non-normal data into the wrong model. GLMs let you analyse your data honestly, report results rigorously, and hold your own in peer review. The course is hands-on, biology-focused, and designed for people who didn't sign up for a statistics degree — but need to do statistics well." If you were convinced by the LLM's sales pitch, here are some details: Date/Time: Tuesday 5th May, from 09:00 to 13:00. Mandatory pre-requisite: Good basic statistics skills. Recommended but not essential: Familiarity with least-squares regression techniques. Cost: Free of charge. How to register: First of all: If you register and then don't attend, then you take away a place from a colleague. Please plan ahead responsibly. 1) Check your calendar! 2) Ask your PI. 3) Go to the Training schedule page https://training.vbcf.ac.at/training/schedule.php , click on the "Register" button for the course and fill out the little form. Thank you, András

April 29, 2026

[VBC-SCIENCE] Statistical Neurodivergence, or: Fitting non-Normal models

by Aszodi,Andras

In our experiments we often count how many times something happened: how many mice died, how many cells got transfected, etc. Such data have non-Normal distributions and can be modelled with "generalised linear models" (GLMs), which is what this course is about. Topics: - How many mice will die of poisoning? Fitting dose-response curves with a binomial GLM (aka "logistic regression") - How many people will die of old age? Mortality rates and the Poisson GLM - The Ebola epidemic: Quasi-Poisson and negative binomial GLM - Back to the roots: Ordinary Least Squares (OLS) regression as a special case of GLM Date/Time: Tuesday 5th May, from 09:00 to 13:00. Mandatory pre-requisite: Good basic statistics skills. Recommended but not essential: Familiarity with least-squares regression techniques. Cost: Free of charge. How to register: First of all: If you register and then don't attend, then you take away a place from a colleague. Please plan ahead responsibly. 1) Check your calendar! 2) Ask your PI. 3) Go to the Training schedule page, click on the "Register" button for the course and fill out the little form. You will get an automatic acknowledgement by email, and an official confirmation from me later next week. Thanks, András

April 24, 2026

[VBC-SCIENCE] Curve fitting course announcement

by Aszodi,Andras

Dear Colleagues, "This is not my AI course", not yet. But if you want to understand neural networks, you need to learn about nonlinear regression first. TOPICS - How to fit a PCR curve? Simple nonlinear regression. - Did my enzyme inhibitor work? Statistical comparison of several nonlinear regression models. - More than one independent variable? Multivariable nonlinear regression. - Non-parametric techniques: LOESS, kernel smoothing methods, splines, generalized additive models. More details at https://training.vbcf.ac.at/training/rnonlinreg.php When? 24th April (Friday), from 09:00 to 13:00. Where? Physically at the IMP seminar room E-016. Non-VBC colleagues may attend online if needed. Mandatory prerequisites: Basic statistics knowledge: "Think Statistics! with R" course https://training.vbcf.ac.at/training/rstat.php or equivalent. Recommended: Familiarity with linear regression ( https://training.vbcf.ac.at/training/rlinreg.php ). How much? Free of charge. How to register? Check your calendar. Then head to https://training.vbcf.ac.at/training/schedule.php , click on the corresponding Register button and fill out the little form. Make sure your PI agrees to your participation. Please keep in mind that registration involves a commitment to attend. General rules: Review them carefully at https://training.vbcf.ac.at/training/practical_information.php . Best regards, András

April 14, 2026

[VBC-SCIENCE] ANOVA course announcement

by Aszodi,Andras

Dear Colleagues, In scientific research we often compare the means of groups of observations. If you have never worked with data like this: then it's just a matter of time and you definitely will :-) This is why I am offering my "ANOVA with R" course which will teach you how to analyse such datasets. TOPICS (see ANOVA course <https://training.vbcf.ac.at/training/ranova.php> for details): The intuition behind ANOVA: Comparing the means of several samples by analyzing variances. Technical details: prerequisites, omnibus F-test, post hoc tests. Getting the sample size right: Power calculations. The relationship between ANOVA and linear regression [optional]. Combination of effects: two-way ANOVA. "Are these regression lines different?" ANCOVA [if we have time]. DATE/TIME: Friday the 17th April, 09:00 - 13:00. LOCATION: "Live" at the IMP seminar room E-016 by default. Online participation via Zoom is available only as a fallback option. PREREQUISITES: Familiarity with R and basic statistics knowledge. Attending my Linear Regression course <https://training.vbcf.ac.at/training/rlinreg.php> which is scheduled on Thursday 16th April is highly recommended as the two methodologies are intimately related. COST: Free of charge. HOW TO REGISTER: First of all: If you register and then don't attend, then you take away a place from a colleague. Please plan ahead responsibly. 1) Check your calendar! 2) Ask your PI. 3) Go to the Training schedule <https://training.vbcf.ac.at/training/schedule.php> page, click on the "Register" button for the ANOVA course and fill out the little form. Note that the linear regression course registration is still open, please do not mix up the two! (But it's all right to attend both... :-) ) You will get an automatic acknowledgement by email, and an official confirmation from me later this week. Thanks, András

April 7, 2026

[VBC-SCIENCE] Linear regression course announcement

by Aszodi,Andras

Dear Colleagues, * The "LLM Castle of AI" (vibe-painted by me personally, with Mr Claude's help) is based on transformer networks. * Transformer networks are neural networks. * Neural networks are nonlinear regression methods. * Nonlinear regression is based on linear regression. This means that linear regression is the cellar of the "AI castle". Join me in exploring this cellar, which harbours no monsters or expensive wine, but some interesting topics, such as: - How to fit straight lines? Single, weighted and multivariable linear regression. - How to model noise in both variables? Orthogonal regression, Principal Components Analysis. - How to simplify the models? LASSO, Ridge, Elastic Net regressions. - What to do with correlated independent variables? Principal component regression, orthogonal polynomial regression. More details at https://training.vbcf.ac.at/training/rlinreg.php WHEN? 16th April (Thursday), from 09:00 to 13:00. WHERE? Physically at the IMP seminar room E-016. Non-VBC colleagues may attend online if needed. PREREQUISITES: Basic statistics knowledge. If you have attended my "Think Statistics! with R" course ( https://training.vbcf.ac.at/training/rstat.php ) then you are well prepared. Some very basic familiarity with R is also required. HOW MUCH? Free of charge. HOW TO REGISTER? Check your calendar. Then head to https://training.vbcf.ac.at/training/schedule.php , click on the corresponding Register button and fill out the little form. Make sure your PI agrees to your participation in the course. Please keep in mind that registration involves a commitment to attend. GENERAL RULES: Review them carefully at https://training.vbcf.ac.at/training/practical_information.php . Best regards, András

April 1, 2026