Mihaly Badonyi

Grateful to @mpi-cbg.de @csbdresden.bsky.social @mpipks.bsky.social for a fantastic work environment (and my ELBE fellowship).

Based on this, we introduce Pinc (“probability of interface native contacts”): a simple and interpretable protein-protein interaction score for #alphafold (AF2/AF3) models.

Colab + CLI: git.mpi-cbg.de/tothpetroczy...

Feedback is appreciated.
#rstats

Two-panel calibration plot (two benchmark dimer datasets) comparing predicted interchain contact-probability bins (x-axis) with the observed fraction of native interfacial contacts (y-axis). Points follow the diagonal, indicating close agreement between predicted probabilities and true interface-contact fractions.

My first manuscript in MPI colours! With @tothpetroczylab.bsky.social, we show that AlphaFold PAE-derived contact probabilities are well calibrated to the fraction of true interface contacts across experimentally determined protein dimers.

www.biorxiv.org/content/10.6...

SS18::SSX activates Polycomb target genes without BAF ❌
Instead, transcription relies on EP300 via the SS18 QPGY domain
www.biorxiv.org/content/10.6...
➡️ Coactivator targeting emerges as a new therapeutic strategy in synovial sarcoma 🎯
Team work from @banitolab.bsky.social and @uoe-igc.bsky.social

The abstract deadline for Mutations in Time and Space 2026 closes in 3 weeks time. Sign up here: coursesandconferences.wellcomeconnectingscience.org/event/mutati...

As well as thrilling science, the programme includes dinner at the beautiful King's College in Cambridge.

@eventswcs.bsky.social

A structure-guided approach to noncoding variant evaluation for transcription factor binding using AlphaFold 3 Abstract. Noncoding single-nucleotide variants (SNVs) that alter transcription factor (TF) binding can affect gene expression and contribute to disease. Se

Our first foray into non-coding variation: structure-guided TF-DNA modelling with AlphaFold 3. Not a replacement for sequence-based predictors, but a complementary way to reason about mechanism. Nice collab with @simonbiddie.bsky.social academic.oup.com/nar/article/...

Can MAVEs and population-free VEPs be combined to improve variant classification? VEPs detect a broad range of pathogenic variants, while MAVEs give more conservative & decisive calls. Combined, they equitably reclassify >90% of VUS. Read more in our study on combining evidence from MAVEs and VEPs:

EASTBIO - Dissecting CIC::DUX4 oncogenic circuitry through single-cell perturbation profiling and network inference at University of Edinburgh on FindAPhD.com PhD Project - EASTBIO - Dissecting CIC::DUX4 oncogenic circuitry through single-cell perturbation profiling and network inference at University of Edinburgh, listed on FindAPhD.com

We have an exciting PhD opportunity through the EASTBIO programme, co-supervised with Diego Oyarzún. This project combines synthetic and systems biology to uncover the gene-regulatory circuitry hijacked in a cancer model. We’re looking for candidates with a strong interest in functional genomics.

acmgscaler: an R package and Colab for standardized gene-level variant effect score calibration within the ACMG/AMP framework academic.oup.com/bioinformati... 🧬🖥️🧪 github.com/badonyi/acmg... #Rstats

Google Colab

3/3 Our study also introduces a simple structural metric, SPDV, proposed as an orthogonal strategy to existing computational tools for aiding the diagnosis of RyR1-related diseases. If you are interested in applying this metric to other proteins, check out Lukas’ Colab notebook:
💻 edin.ac/40yke9l

Google Colab

2/3 If you work on MAVEs, or are a clinician curious about how strong MAVE or VEP evidence is in a gene, check out 𝚊𝚌𝚖𝚐𝚜𝚌𝚊𝚕𝚎𝚛. It calibrates functional scores to #ACMG/AMP standard, and you can run it via this Colab notebook:
💻 colab.research.google.com/github/badon...

1/3 In this work on RyR1, led by Rolando, we (@marshlab.bsky.social) highlight the limitations of using ROC AUC alone to assess clinical utility. Future approaches should consider classification behaviour across the full score distribution.
📄 onlinelibrary.wiley.com/doi/epdf/10....

I'm stoked to be organising next year's MSS right here in beautiful Melbourne! We know Australia is very far away, and we're working hard to make sure we can support as many ECRs to attend as possible, so please do register and apply for a travel award!

Thanks also to @natcomms.nature.com for the smooth review process!

8/8 Our work raises many interesting questions, but I will leave you with one: when a disease can be caused by both dominant and recessive mutations, how can we predict which variants are pathogenic only in the recessive state?

Google Colab

7/8 Finally, we made it easy to compute mLOF for a set of missense variants: there is a user-friendly Google Colab notebook available for anyone to try.
💻 edin.ac/48Aysep

6/8 We also find that many genes can cause different diseases through different mechanisms. This carries an important message for therapeutic development, suggesting that different phenotypes of the same gene may need distinct treatment strategies.

5/8 Therefore, we applied mLOF to 2,837 phenotypes across 1,979 Mendelian disease genes. In dominant genes, nearly half of phenotypes appear to arise from non-LOF mechanisms.

4/8 Perhaps the most powerful validation of mLOF lies in multiplexed assays of variant effect #MAVE data. LOF, DN, GOF, and hyper-complementing variants identified by MAVEs are distinguished consistently and effectively by the mLOF score.

3/8 We introduce a protein structure-based missense LOF (mLOF) likelihood score, which predicts if mutations lead to LOF based on how much they cluster in structures and how energetically destabilising they are.

2/8 Knowledge of mechanisms is hugely important for treating genetic diseases: LOF mutations can usually be treated with gene replacement, but other mechanisms often require more targeted approaches.

1/8 Our new paper in Nature Communications explores how often pathogenic missense variants cause disease through loss-of-function (LOF), gain-of-function (GOF), or dominant-negative (DN) effects.
📄 nature.com/articles/s41...

Happy to share that 𝚊𝚌𝚖𝚐𝚜𝚌𝚊𝚕𝚎𝚛 is now on CRAN! 🎉
This means long-term stability and easy installation with:
𝚒𝚗𝚜𝚝𝚊𝚕𝚕.𝚙𝚊𝚌𝚔𝚊𝚐𝚎𝚜('𝚊𝚌𝚖𝚐𝚜𝚌𝚊𝚕𝚎𝚛')

🗞️ doi.org/10.1093/bioi...

#rstats #acmg #varianteffect #MAVEs #VEPs #genomics

Yes, if vaccines count as a discovery, I would add the Haber-Bosch process for producing ammonia, and by extension, fertiliser to feed the world.

acmgscaler: An R package and Colab for standardised gene-level variant effect score calibration within the ACMG/AMP framework A genome-wide variant effect calibration method was recently developed under the guidelines of the American College of Medical Genetics and Genomics and the Association for Molecular Pathology (ACMG/A...

We’ve updated the acmgscaler manuscript following reviewer and community feedback.

The R package now has a single calibrate() function, and the Colab interface is easier to use.

📄 Manuscript: www.biorxiv.org/content/10.1...
🧪 Colab: edin.ac/4mjzijp

#rstats @theacmg.bsky.social

📌

In the previous version, I found the functionality where hovering over the sequence highlighted the corresponding residue in the structure to be very useful. Any chance this could be brought back while keeping the `copy sequence` button?

Why variant effect predictors and multiplexed assays agree and disagree https://www.biorxiv.org/content/10.1101/2025.07.31.667868v1

Disease-associated genetic variants can cause mutations in tissue-specific protein isoforms Genetic variants can cause protein-coding mutations that result in disease. Variants are typically interpreted using the reference transcript for a gene. However, most human multi-exon genes encode al...

Congratulations to @gweykopf.bsky.social for her first first author manuscript, now on biorxiv.

www.biorxiv.org/content/10.1...

Many thanks to all involved - @wbickmor.bsky.social @mbadonyi.bsky.social @eliasfriman.bsky.social, Joe Marsh, Jasmine Nguyen, Mark Gorrell and others.

A knowledge-based distance metric highlights underperformance of variant effect predictors on gain-of-function missense variants https://www.biorxiv.org/content/10.1101/2025.07.23.666325v1