Lili (@LILILI_LLL) header
Lili (@LILILI_LLL) profile picture
Joined September 2020
Photo by Lili (@LILILI_LLL)
📢Big announcement 📢 Today we release μSAM2 for interactive and automatic microscopy segmentation, classification, and tracking. μSAM2 was trained on > 30M annotated cells, nuclei and neurites in 2D and 3D light and electron microscopy. It enables SOTA 2D/3D segmentation ...
5
43
1
188
9,217
Online Now: Condensates increase the frequency of transcription dlvr.​it/TWM5sB
17
73
5,356
We ran a similar experiment a few months ago. Back then, when we asked an LLM to predict a protein structure on its own, it mostly made things up. We spent a lot of time thinking about how to improve native llm folding through post-training. Now it feels like that gap is already close to closing😢 That’s a little scaring. I used to think of AlphaFold as a form of narrow superintelligence. Even a team of top computational biologists would struggle to reason their way from a sequence to an accurate structure. If a general-purpose LLM can reach that level without calling any external tools, how many other domains could it eventually outperform human experts in? maybe we’re already seeing the signs of general superintelligence.
Given Claude Fable 5.2 can predict protein sequence, I was curious if it can predict structure. I gave it a relatively simple prompt: "predict contacts, 2ndary struct, use dynamic prog. and linear alg. to predict structure" It was able to predict fold of ubiquitin correctly 🤯
6
8
2
117
7,924
A 2023 Review in @ScienceTM looked at the complex connections between astrocytes and other types of cells in the nervous system, including neurons, oligodendrocytes, and microglia. Learn more: scim.ag/3USPZdx #ScienceMagArchives
8
67
2
281
25,396
Lili retweeted
Cellpose-SAM/Cellpose4: superhuman generalization for cellular segmentation github.com/mouseland/cellpos… vs Cellpose3 CellSAM microSAM SAMCell PathoSAM Omniposeac @marius10p @computingnature bioRxiv 2025 biorxiv.org/content/10.1101/…
2
34
192
11,605
👀What if you could record calcium dynamics in as many cells as you want simultaneously, and read the temporal records in situ with a single endpoint imaging session? 🚀Introducing TCaMP, a genetically encoded calcium recorder. 🧬TCaMP is a new member of the CytoSense family, and writes Ca²⁺ dynamics along a protein tape (CytoTape) that elongates inside the living cell. Time becomes position along the tape. Like tree rings :deciduous_tree:, but for Ca²⁺ dynamics. 〽️The record is analog and continuous: Ca²⁺ levels on a single intensity scale along a continuous time axis, for scalable endpoint readout. 🔗TCaMP is built on split-1POK, a split tape monomer scaffold whose two halves reconstitute when the domains fused to them interact. Split-1POK is programmable: swap the interaction pair fused to it, and the same scaffold records that pair’s interaction dynamics across time instead. 🔬In human cells, TCaMP recorded Ca²⁺ from transmembrane entry, receptor activation, and intracellular store release, and detected stimulations as brief as 10 minutes. ⏱️At hour-scale kinetics over days, TCaMP complements fluorescent Ca²⁺ indicators, trading sub-second temporal resolution for scalability and recording duration. We envision TCaMP and the general split tape monomer strategy being useful for questions that span large cell populations across hours to days and beyond. 👏Work by Yixiao Yan, a PhD student in the lab. 📄 Preprint: biorxiv.org/content/10.64898…
3
43
3
154
12,109
It's quite rare to be able to photograph a winter plumage Puffin especially inland, so had to take the opportunity today. Tees Barrage this morning
3
6
52
957
FlAsH-ID enables live-cell photoproximity labeling with a short peptide tag nature.com/articles/s41589-0…
1
16
1
68
9,030
Lili retweeted
🧠 How do brain cells maintain their identities and functions for decades? Thrilled to share our new study in Cell! We mapped the regulatory genome of the adult human brain to explore cell identity, developmental memory, and disease risk: doi.org/10.1016/j.cell.2026.… 🧵 (1/14)
2
45
2
178
13,892
Behold: The full JUMP Cell Painting chemical dataset preprint! Images + 1.53 billion single-cell profiles for 115,796 compounds in human cells, made across 12 labs. All freely available 🧵 biorxiv.org/content/10.1101/…
13
59
5
244
14,166
The blood-brain barrier doesn't just keep things out. In @CellCellPress, 92 of 861 plasma proteins entered the mouse brain. In early 5xFAD mice, APOA1 uptake fell; boosting liver APOA1 reduced astrocyte lipid load. Kudos to Haoyue Zhou & @andcyang 👏 doi.org/10.1016/j.cell.2026.…
1
14
1
50
3,275
We wrote a biological light sheet into the brain🤯 SPress confines expression to a thin chosen plane for kilohertz, large-scale voltage imaging with a standard widefield microscope. Optical sectioning, encoded through gene expression. Preprint: doi.org/10.64898/2026.10.30.…
5
53
8
196
20,839
Online now: An astrocyte-microglia MFG-E8 axis promotes synapse engulfment in early amyloid pathology dlvr.​it/TWG7SR
10
42
4,022
耶鲁计算机博士生刘晨,把自己发在 Nature Machine Intelligence、NeurIPS 等处的论文配图脚本整理成仓库,后来又做成了 AI Skill。 现在 GitHub 一万多颗星。 写论文要画柱状图、雷达图,可以让 AI 照着画。 github.com/ChenLiu-1996/figu…
41
285
11
1,790
132,071