Vermeer: Autoregressive generative modeling of microscopy predicts protein localization

bioRxiv · 12 Nov 2026 · Preprint · Open access

Kambhampati, Zimmermann + 4 more

The abstract is read at the publisher; the summary is JClub's.

Vermeer, a channel-adaptive autoregressive generative model trained on the Human Protein Atlas, generates microscopy images of protein localization with substantially improved perceptual quality and biological fidelity.

Why it matters: Imaging human proteins across all factors affecting localization is experimentally infeasible, limiting scalable modeling of where proteins reside in cells.
What they did: Vermeer conditions generation on protein sequences and landmark stains, and uses an autoregressive framework to support different channel subsets/orderings and zero-shot transfer across imaging conditions.
The result: The model generalizes to unseen proteins and cell lines, and its flexible channel handling enables transfer to datasets with different channel configurations than training, supporting scalable protein-localization modeling.

AI summary of the abstract · may contain errors · not medical advice · read the paper.

Open access
1 cites

Free. We’ll email you when a new paper cites this one or any other paper you follow on JClub — at most one email a day. Every email has a one-tap unsubscribe.

Robust Cable Models for Complex Neuronal Morphologies
bioRxiv · 12 Nov 2026 · Preprint
AI summary · may contain errors
MASCAF fits cable models to closed 3D surface mesh neuronal morphologies with cyclic topology, enabling robust modeling of barn owl auditory “toric spines” despite topological holes.
- Open access
Open
Share
Post on X Post on Bluesky Share on LinkedIn
https://jclub.com.au/a/10.64898/2026.05.10.721501
Open in JClub
GraphHDBSCAN*: Graph-based Hierarchical Clustering on High-dimensional Spaces with Applications to Single-cell RNA Sequencing Data
bioRxiv · 11 Nov 2026 · Preprint
AI summary · may contain errors
GraphHDBSCAN* recovers biologically meaningful hierarchical cell populations from scRNA-seq data, matching curated references more closely and resolving monocyte subpopulations.
- Open access
- 2 cites
Open
Share
Post on X Post on Bluesky Share on LinkedIn
https://jclub.com.au/a/10.64898/2026.03.24.713924
Open in JClub
Quantitative profiling of whole-brain projectomes at single-axon resolution using deep learning and high-resolution light sheet microscopy
bioRxiv · 10 Nov 2026 · Preprint
AI summary · may contain errors
MAPL3, an end-to-end deep learning pipeline for tera-voxel-scale light-sheet fluorescence microscopy, reconstructs brain-wide axon projections at single-axon resolution and generalizes across tracers, microscopes, and protocols.
- Open access
- 3 cites
Open
Share
Post on X Post on Bluesky Share on LinkedIn
https://jclub.com.au/a/10.1101/2025.11.14.688340
Open in JClub

Journals this month

All journals →

Topics moving, week to 13 Nov 2026

All topics →

By field

All journals by field →

Vermeer: Autoregressive generative modeling of microscopy predicts protein localization

Related in JClub

Robust Cable Models for Complex Neuronal Morphologies

GraphHDBSCAN*: Graph-based Hierarchical Clustering on High-dimensional Spaces with Applications to Single-cell RNA Sequencing Data

Quantitative profiling of whole-brain projectomes at single-axon resolution using deep learning and high-resolution light sheet microscopy

Browse JClub