iiinova — where your digital self keeps evolving

Rafael Irizarry @rafalab.bsky.social · 18/05/2026

A flaw in Person-Δ may be overstating progress in single-cell perturbation prediction models. Pearson warned about this in the 19th century: reusing the same controls induces spurious correlation. Split the controls, and much of the claimed prediction power fades: www.biorxiv.org/content/10.6...

biorxiv.org

0146

Rafael Irizarry @rafalab.bsky.social · 01/12/2025

📢 We are taking applications for our Postdoctoral Fellows Program at Harvard/DFCI! 🔹Join a research group in our department 🔹Co-mentoring opportunities with 2+ faculty 🔹Collaborate with investigators beyond our department 🔹Salary starts at $75K Apply here: t.co/B7SLZzQFKu

t.co

https://ds.dfci.harvard.edu/postdocs/

0915

Rafael Irizarry @rafalab.bsky.social · 24/12/2024

We agree. I meant function in its most basic definition: every p-dimensional x in your dataset is mapped to a unique 2-dimensional f(x). I did not claim f is defined for all p-dimensional xs. My point is that this f is difficult or impossible to describe. In contrast, we can write it down for PCA.

010

Rafael Irizarry @rafalab.bsky.social · 24/12/2024

I like it. Thank for the tip.

000

Rafael Irizarry @rafalab.bsky.social · 24/12/2024

I agree. I don't use them at all with genomics data, especially sparse noisy scRNA-Seq data. It does appear to perform impressively well with high signal-to-noise ratio datasets, such as MNIST.

110

Rafael Irizarry @rafalab.bsky.social · 24/12/2024

It depends on the specific biological insight you want to highlight or communicate.

020

Rafael Irizarry @rafalab.bsky.social · 24/12/2024

The inertia here is strong. I have not been unable to convince collaborators to not use them on papers I am a co-author on... I'll keep trying though. At some point I might give up as I did with pie charts: simplystatistics.org/posts/2012-1...

simplystatistics.org

Simply Statistics: I give up, I am embracing pie charts

130

Rafael Irizarry @rafalab.bsky.social · 24/12/2024

As made clear in the blogpost I am not against UMAP either. But when I see a plot in a paper, I want to understand what I am being shown and why. Other than to show different cell types have different expression patterns, which I already know, or to decorate, why use UMAP to display in 2D?

362

Rafael Irizarry @rafalab.bsky.social · 24/12/2024

What am I supposed to learn from that plot? Cell types have different expression patterns. Those are markers for different cell types. So, this just confirms something obvious. Do all those non-linear shapes and tiny clusters represent anything biological?

240

Rafael Irizarry @rafalab.bsky.social · 23/12/2024

To be clear, as the post explains, UMAP can be useful for exploratory data analysis. My concern is their inclusion in papers as if they were results. What exactly is the reader supposed to learn? And how often are we misdirected by false clusters or artifactual shapes?

2151

Rafael Irizarry @rafalab.bsky.social · 23/12/2024

Can you explain what the axes represent? As mentioned in the post, UMAP can be useful for exploring data. But why are plots included in papers? What is the reader supposed to get out of them? The 2D distance between points can't be interpreted. It seems the only reason is because they are pretty.

7100

Rafael Irizarry @rafalab.bsky.social · 23/12/2024

There are plenty of alternatives. They don’t produce flashy art work but they do provide scientific insights. If journals want art work no need to pretend we are analyzing data. Just paint pretty pictures.

190

Rafael Irizarry @rafalab.bsky.social · 23/12/2024

This is unfortunately true. I would say the main reasons are that the subject is hard and deep understanding is not incentivized enough. But note understanding UMAP is much harder than understanding pvalues.

030

Rafael Irizarry @rafalab.bsky.social · 26/11/2024

To apply use links below: 1️⃣ Tenure-track (any rank) in AI/ML academicpositions.harvard.edu/postings/14387 2️⃣ Assistant Professor in Single Cell Genomics academicpositions.harvard.edu/postings/14416 3️⃣ Lecturer & Director Training/Education careers.dana-farber.org/job/9572/dir...

academicpositions.harvard.edu

Assistant/Associate/Full Professor of Data Science and Biostatistics

The Departments of Biostatistics at Harvard T.H. Chan School of Public Health and Data Science at the Dana-Farber Cancer Institute provide exceptional environments to pursue research and education in ...

110

Rafael Irizarry @rafalab.bsky.social · 26/11/2024

🚨 We're hiring! 🚨 Join us at @HarvardBiostats and @dfcidatascience We are filling THREE positions: 1️⃣ Tenure-track (any rank) in AI/ML 2️⃣ Assistant Professor in Single Cell Genomics 3️⃣ Lecturer & Director Training/Education Need to apply separately for each role. Links 👇

151

Rafael Irizarry @rafalab.bsky.social · 22/11/2024

New method for identifying spatially variable genes in spatial transcriptomics. We use spectral graph theory to extract 1D curves related to tissue morphology, creating a more relevant coordinate system. Improves sensitivity. with @PhillipNicol et al. www.biorxiv.org/content/10.1...

04111

Rafael Irizarry @rafalab.bsky.social · 07/12/2023

Question for #rstats educators: Do you use a standard format for exercises? I have 100s of exercises along with solutions in quarto documents (see image). But they are not formatted in a systematic way. I would like to easily create docs with and w/o solutions and export to LMS.

150

Rafael Irizarry @rafalab.bsky.social · 19/10/2023

Hoy inicia nuestro Curso en español de Ciencia de Datos: Fundamentos de R en HarvardX! Aprenderás desde los básicos hasta data wrangling, conceptos de programación e importación de datos. La versión gratis y en línea te da acceso a los videos, textos y ejercicios. t.co/TGID4IwTSQ

0104

Rafael Irizarry @rafalab.bsky.social · 22/09/2023

Data visualization is used in two ways in data analysis: 1) Communicating data insights. 2) Guiding analysis and spotting data issues. While modern courses and books focus on the first, it's the second - exploratory data analysis - that is more important and harder to teach.

172

Rafael Irizarry @rafalab.bsky.social · 20/09/2023

Introducing the Data Science Postdoctoral Fellows Program at Harvard/DFCI! - Join a research group in our department - Co-mentoring opportunities with 2+ faculty - Collaborate with DFCI investigators beyond our department - Salary starts at $75K Apply here: ds.dfci.harvard.edu/postdocs/

035