Yang Zhang

[Yin Shen @YinShenLab]

I’m delighted to share our latest work on the 3D epigenome of glial cell types in the developing human cortex, published today in Nature!
https://t.co/wB1JDj0F2Z

[Jian Ma @jmuiuc]

very comprehensive Review on single-cell foundation models from @JiayuanDing, @Xiaojie_Qiu, @TheodorisLab, and colleagues. After going over it, I would encourage the authors to make it more critical rather than primarily an inventory of models.

First, I think Fig 1 needs some rethinking. The definition of scFM has become too loose. For example, the actual model unit of GET is celltype pseudobulk aggregated from scATAC-seq, hence not single cell model. Also thanks for highlighting our TissueNarrator, but we certainly do not consider it a bona fide scFM - it is a specialized spatial transcriptomics model built by leverging an existing foundation LLM (Qwen). TissueNarrator is very interesting and we re having lots of fun w/ it but it's not a scFM. Anyway - Fig 1 seems to mix FMs, specialized models built on FMs, and task-specific models into one big family tree.

More broadly, I think the more interesting question for a scFM review in 2026 is not how many scFMs now exist or how their architectures differ, but what has the scFM paradigm actually bought us? Since the first wave around scBERT/Geneformer/scGPT, the number of models, parameters, and pretraining cells has exploded, but as a field we are not convinced that capabilities have advanced proportionally.
- Where has large scale pretraining enabled something that a strong specialized model could not do ?
- Where do scFMs consistently beat strong task specific or even simple baselines?
- Where is the broad evidence for true cross-context generalization on biologically interesting tasks?

The Review mentions many of these challenges indeed, but I think they should be much closer to the central theme of the article rather than appearing mainly in a later Challenges section :-)

I might almost suggest to flip the review around and instead ask:
- what have we actually learned after several years and many scFMs?
- What claims have survived independent benchmarking?
- What has scaled and what has saturated?
- What capabilities are genuinely attributable to pretraining and where do we still need much more evidnece?
- What important biological problems remain essentially unsolved by scFMs ?

Given the current rate of model proliferation, there may be another 50 scFMs by this time next year and this inventory / figure will become outdated very quickly. To me, a more useful and timely "state of the field" critique would be much more valuable and long-lasting than another "model zoo" review .

[evo-devo @Xiaojie_Qiu]

Are single-cell foundation models (scFMs) a true biological breakthrough, or just over-hyped architecture tweaks?

To separate genuine conceptual advances from incremental hype, the lead authors of pioneer models, including Geneformer, Cell2sentence @david_van_dijk , Nicheformer @fabian_theis , GET @raulrabadan, CellPLM @tangjiliang , and our Tabula, have united to publish a systematic, critical synthesis of the scFM landscape.

Designed for the broader scientific community, this guide cuts through the noise to show how emerging model designs can truly support grounded biological discovery.

📄 Paper: https://t.co/FE6TloCWqs
📷 Curated paper list: https://t.co/pCdWf45OjK

Led by the incredible @JiayuanDing

and @shiyu_jiang23 , Zhaoyu Fang, Yujie Zhang, @xutzhang , @jkobject , Weixu Wang, @alexanderfuxi , Aakash Patel, Syed Rizvi, @Y_Ryan_Lu , @SiyuHe7 , @YixinxinWang , @KejunYing , @peterpaohuang , @YifanLu2024 , @Nanguage , Mengchen Wang, Ziyang Miao, Jianhui Lin, Jimmy Ding, Jerry Wang, @imweio , @TianlongChen4 , Guoxian Yu, Min Li, Jiayi Ma, @feiwang03 , @Yuyingxie , @cmuptx , @PengHeAtlas , Emily B. Fox, @dasongle , @ericxing

Give it a read, we hope you will enjoy it!

[CMU Kenneth C. Griffin School of Computer Science @GriffinCMU]

In a paper published in Science, computational biology researchers at Carnegie Mellon University shed new light on Alzheimer's disease that could lead to improved treatment options.
https://t.co/vZlaWSGwSB

[Jian Ma @jmuiuc]

As a collaboration between my group at @SCSatCMU, Aviv Regev's team at @genentech, and @michellearning's team at @medra_ai, we are pleased to share our Perspective, "Towards human-led, agent-driven autonomous laboratories for the life sciences." 🤖 https://t.co/tJBVnIXtTr

This preprint release is coordinated with a complementary Perspective by @MengdiWang10, @lecong, and colleagues: https://t.co/x2AFNMmEkC. Together, the two Perspectives examine the future of autonomous science from complementary angles.

Our Perspective was Led by @WenduoC in my group, with @mishamamq, @ShenShuaik4260 and @zocean636 at CMU; Anna Hupalowska, Jennifer Rood, Christine Bakan, and Aviv Regev at @genentech/@Roche; and Gaurav Agrawal and @michellearning at @medra_ai.

🚀https://t.co/tJBVnIXtTr

[Jian Ma @jmuiuc]

Very excited to share our @ScienceMagazine paper on single-cell #3D #genome reorganization in #Alzheimer's disease.
We jointly measured gene expression and 3D genome architecture in individual human brain cells using #GAGEseq, then integrated these data w/ chromatin accessibility and spatial transcriptomics. We uncovered increased #compartment #mingling and distance-dependent rewiring of gene regulatory contacts in AD.
We also developed #Hicformer, a transformer-based model that integrates DNA sequence with 3D genome features to predict cell type-specific gene expression and prioritize candidate regulatory elements.
Huge kudos to co-first authors @zocean636 and @xinyuelu1999; and many thanks to Zhijun Duan @UW, Hansruedi Mathys @PittTweet, & David Bennett @rushalzheimers for the wonderful collaboration, as well as to all our co-authors. @CarnegieMellon @SCSatCMU @CMUCompBio #AlzheimersDisease #3DGenome #SingleCell #AI

https://t.co/G3yPdRMeym

[Nature Methods @naturemethods]

HiCFoundation: a foundation model trained on Hi-C data for comprehensive 3D genome and epigenomics analysis.

https://t.co/PFqHNWukVZ

[Anshul Kundaje @anshulkundaje]

The @ENCODE_NIH Phase 4 Registry of candidate cis regulatory elements (ccREs) (2.37 million human and 967,000 mouse cCREs) is officially out

https://t.co/uuJAsO334Z

https://t.co/wtCW3beYlv

Congrats to @MooreJillE
@ZhipingWeng & everyone else.

Lots more incoming ...

[Jian Ma @jmuiuc]

Happy to share #DNALONGBENCH @NatureComms. We challenge DNA #FoundationModels w/ long-range sequence context and hope this sparks more meaningful ways to evaluate growing AI+BIO models. Kudos to @WenduoC @ZhenqiaoSong @zocean636. Great collab w/ @lileics. https://t.co/kDGUvKAL1I

[Qian Cheng @sgqcheng]

DNALONGBENCH: a benchmark suite for long-range DNA prediction #DNAfoundationmodels #BiotechNatureComms
@WenduoC
@ZhenqiaoSong
@zocean636
@lileics 
@jmuiuc  

https://t.co/xvafxE1jNY

[Jian Ma @jmuiuc]

Final version of our paper @eLife (led by @belmont_andrew): Major nuclear locales define nuclear genome organization and function beyond A and B compartments.
https://t.co/Nw6eSDW2Pi

[Nature Methods @naturemethods]

A new benchmarking study assesses methods for comparing chromatin contact maps in 3D genome research. @GjoniKatie, @EvonneMcArthur, @GladstoneInst

https://t.co/9kk8mmSb3H

[Jian Ma @jmuiuc]

Proper and meaningful benchmark datasets are crucial for advancing genomic LLMs/FMs, and ML methods for genomics in general. Fantastic collab w/ @lileics's group. Amazing work led by @WenduoC @ZhenqiaoSong @zocean636

[Andrew Belmont @belmont_andrew]

Genome-wide analyses correlate nuclear speckle proximity with gene expression. We share our minireview discussing possible functional relationships between nuclear speckles and gene expression in both health and disease: https://t.co/csVBLeRCO3

[James Zou @james_y_zou]

Very excited to share our primer on LLM for biology @naturemethods! We discuss how to use:
⚡️human & protein LMs (eg #ESM)
⚡️#singlecell & #multimodel LMs
⚡️when to fine-tune, transfer learn etc.
Paper https://t.co/V2YswS1A1K
Colab tutorial https://t.co/xsDcqWJPrp
Written for a broad audience. Great job @ElanaPearl @KyleWSwanson!

[Michael Kosicki @michaelkosicki]

Beta-version of the new VISTA Enhancer Browser, 7 years in the making, with support for developmental stages, different species and variant comparison is finally online. Feedback is welcome! https://t.co/2y5HDCYaKD #enhancers #embryos #evodevo

[Yang Zhang @zocean636]

@cliao345 @GeiselMed Congrats! That's amazing news!

[Xuegong @XuegongZ]

Our scFoundation has been finally published. It is currently one of the largest cellular models, pretrained on the transcriptomes of 50M single cells. Check out its performance on all the downstream tasks. https://t.co/6p8xXZkmGX

[Andrew Belmont @belmont_andrew]

Excited to share new results from our 4DN collaboration between the Gilbert, Ma, van Steensel, and Belmont research groups. Joint analysis, led by our trainees, compared chromosome positioning relative to multiple nuclear locales across 4 cell types
https://t.co/tdO6gpkPbg

[Nature Methods @naturemethods]

scGHOST offers a computational tool to annotate single cell subcompartments from scHi-C or imaging data using graph-based learning.
@jmuiuc, @KyleXiongCMU, @RuochiZhang, @CMUCompBio, @SCSatCMU, @CarnegieMellon

https://t.co/Q1VaUh5FXW