iiinova — where your digital self keeps evolving

Reposted by Chakravarthi Kanduri

Greifflab @victorgreiff.bsky.social · 22/10/2025

📢 We are announcing the Adaptive Immune Profiling Challenge 2025! Can you predict immune state labels from adaptive immune receptor repertoires? 💰 $10,000 prize pool! 🗓️ Launches Nov 5 on @kaggle.com More Info: uio-bmi.github.io/adaptive_imm...

12512

Chakravarthi Kanduri @chakri.bsky.social · 18/08/2025

12/12🙏 Thanks to all collaborators & co-authors for useful inputs, brainstorming and perspectives: Maria Mamica, Emilie Willoch Olstad, Ingrid Hobæk Haff, @manuelazucknick.bsky.social , Jingyi Jessica Li, & Geir Kjetil Sandve.

000

Chakravarthi Kanduri @chakri.bsky.social · 18/08/2025

10/12 As a safer alternative, consider the Benjamini-Yekutieli (BY) method when you can tolerate a bit more type II error. It doesn't completely eliminate the issue but makes these large false positive events much less frequent and severe (a good compromise between BH and FWER).

100

Chakravarthi Kanduri @chakri.bsky.social · 18/08/2025

9/12 Use negative controls/synthetic null data and other diagnostic checks as recommended in the article to identify and minimize caveats. If continuing to use BH method — try to know its assumptions and formal guarantees to ensure correct interpretation of the findings.

100

Chakravarthi Kanduri @chakri.bsky.social · 18/08/2025

8/12 Issues like broken test assumptions, study biases, or the researcher’s flexibility in analyzing the data can make this problem even worse. So, what can you do? We suggest a few key strategies:

100

Chakravarthi Kanduri @chakri.bsky.social · 18/08/2025

7/12 This statistical artefact can lead researchers to incorrectly conclude the existence of an underlying biological mechanism, which might even form the main conclusion of their study.

100

Chakravarthi Kanduri @chakri.bsky.social · 18/08/2025

6/12 It feels intuitive to believe that if hundreds or thousands of features are flagged as significant, at least some of them must be real. However, we show this intuition can be wrong; it's possible that every single finding is false.

100

Chakravarthi Kanduri @chakri.bsky.social · 18/08/2025

5/12 A Counter-Intuitive Trap: Using real-world and simulated data (methylation, gene expression, metabolite and eQTL analyses), we found this phenomenon to be persistent. The primary danger is that researchers may be misled by the sheer volume of these false findings.

110

Chakravarthi Kanduri @chakri.bsky.social · 18/08/2025

4/12 This happens because dependencies in the data can cause many features to falsely appear significant together. While the overall FDR is controlled (e.g., <5% of experiments have errors), the experiments that do have errors can have thousands of them.

100

Chakravarthi Kanduri @chakri.bsky.social · 18/08/2025

3/12 Even when a study has no true biological signal (all null hypotheses are true), the BH method can occasionally generate thousands of statistically "significant" findings.

100

Chakravarthi Kanduri @chakri.bsky.social · 18/08/2025

2/12 The widely used False Discovery Rate (FDR) control method, Benjamini-Hochberg (BH), is a staple in omics research. But when analysing datasets with dependencies between features (like gene expression, methylation, metabolites, QTL analyses ++), it can behave unexpectedly.

100

Reposted by Chakravarthi Kanduri

Puneet Rawat, PhD @puneet021192.bsky.social · 14/12/2024

Happy to share our new collaborative work! 🚨 We analyzed 2250 TCR repertoires to uncover how HLA risk alleles shape immune autoreactivity in T1D. T1D-specific HLA-motifs were also validated in pancreatic lymph nodes (pLN) & spleen. 🧬🧵 #T1D #Immunology #TCR 1/9

1369

Reposted by Chakravarthi Kanduri

Greifflab @victorgreiff.bsky.social · 24/10/2023

Machine learning on immune receptors/repertoires is an exponentially expanding field, but labeled data to benchmark ML methods are missing. We address this need with our new simulation framework LIgO. www.biorxiv.org/content/10.1... Led by @mchernigovskaya.bsky.social and Milena Pavlović.

135