Excited to share our new work on Meta Flow Matching (MFM)! 🎉 MFM helps improve the generalization of personalized treatment response predictions by conditioning on learned features of the initial distribution.
Work led by
@lazar_atan
and
@NZhang211
.
🚀Introducing — Meta Flow Matching (MFM) 🚀
Imagine predicting patient-specific treatment responses for unseen cases or building generative models that adapt across different measures. MFM makes this a reality.
📰Paper:
💻Code:
Excited to share our latest work on Metric Flow Matching (MFM)! Loved working with this team. Amazing to see how far we've come computationally since I started working in this field.
Check out this extension of DynGFN ( with
@lazar_atan
and
@jasonhartford
) which improves scalability with a more efficient parameterization. Reducing trajectory lengths from n^2 --> n.
Curious about how to discover causal interactions among many genes in a cell with uncertainty estimation?
@TrangNguyen2399
@AlexanderTong7
Check our new work :
Diffusion models are dead - long live joint conditional flow matching! 🙃
Tomorrow
@AlexanderTong7
presents his "Improving and generalizing flow-based generative models with minibatch optimal transport"
On Zoom 11am EDT / 3pm UTC:
New paper!🤗
"Harmonic Self-Conditioned Flow Matching for Multi-Ligand Docking and Binding Site Design"
Generating pockets to bind small molecules has applications like designing antidotes that sequester toxins or as a first step toward Enzyme design
1/11
This week, 8/06 @ 4 pm ET, we'll have
@AlexanderTong7
present
Sequence-Augmented SE (3)-Flow Matching For Conditional Protein Backbone Generation
Read about FoldFlow
FF2:
FF1:
Sign up to receive the Zoom link via email!
Thank you
@AlexanderTong7
from
@Mila_Quebec
for having stopped over at
@HelmholtzMunich
and sharing your current work perspectives; Looking forward to your come-back, to enjoy more discussions on cell causality (and the famous Munich beach) (+thanks
@THJimmyLee
for the photo)
It has been such an enormous privilege to work with
@KrishnaswamyLab
and
@AlexanderTong7
on this project.
Alex is currently on the faculty job market and anyone who has any sense should hire him immediately.
Great tutorial on Trellis created by
@Maria_Ramos_Z
which shows how to combine hierarchical clustering and optimal transport for compare a large number of single-cell distributions.
Here it is! We have now created a how-to tutorial to understand and use Trellis, a tool for hierarchical clustering that enables the fast calculation of distances across thousands of distributions containing single-cell information:
Happy to be defending my thesis in a few hours! If you want to hear about our work in graphs, optimal transport, and deep learning, hope to see you there!
📣Two new blog posts - a comprehensive review of Graph and Geometric ML in 2023 with predictions for 2024.
Together with
@mmbronstein
, we asked 30 academic and industrial experts about the most important things happened in their areas and open challenges to be solved.
🧵 1/n
Happy to share that FoldFlow, the first Flow Matching generative model for Protein backbone generation got accepted to
#ICLR2024
as a Spotlight with scores 8/8/8/8! Thanks to the reviewers for the feedback and ofc all my incredible co-authors for pulling this fun paper off🚀!
I will be at
#NeurIPS2023
this week! I am excited to be presenting DynGFN during the main conference. Looking forward to seeing everyone there!
Poster session 4, Wed, Dec 13th 5 - 7 pm CST
@VectorInst
@Mila_Quebec
@valence_ai
1 week left to apply for our computational PhD project 'Single-cell Signalling Analysis of Stem Cell Polarisation in Cancer'.
4-years, fully funded, co-supervised across
@uclcancer
@crick
with a placement at
@Yale
Pls RT!
@alexechu_
@bose_joey
@json_yim
Hi
@alexechu_
, I was also curious about this. (1) It turns out that FM and diffusion have slightly different conditional paths as shown in Figure 3 of . (2) the conditionals for the VE ODE are an optimal transport map for a linear noise schedule.
@KrishnaswamyLab
@CellCellPress
@christophertape
@Maria_Ramos_Z
I’m excited to see what types of methods can be developed using this data. High-level summary, 25m cells, 44 features in a tensor of 10 patients, 11 drug combinations, 3 co-cultures states, with 3 replicates, which presents many opportunities for splitting and validation. (9/n)
@KrishnaswamyLab
@CellCellPress
@christophertape
@Maria_Ramos_Z
We developed Trellis as an extremely fast and interpretable method to understand how a treatment changes the observed cell states. Trellis is able to measure the changes in changes of cell state distributions by how much work it takes to transport over a cell state tree. (6/n)
@KrishnaswamyLab
@CellCellPress
@christophertape
@Maria_Ramos_Z
The L^1 distance between these vectors is now the Wasserstein distance. This means that when used to create manifold-based embeddings between distributions which only require nearest neighbor distributions, Trellis scales log-linearly in the number of cells and samples. (8/n)
@KrishnaswamyLab
@CellCellPress
@christophertape
@Maria_Ramos_Z
Trellis embeds distributions to vectors where the Wasserstein distance between distributions is equivalent to the L^1 distance between vectors. On a tree with hierarchical clustering, we compute the cell abundances in each node of the tree weighted by the edge weight. (7/n)
@KrishnaswamyLab
@CellCellPress
@christophertape
@Maria_Ramos_Z
Since this dataset is so large, it was collected using many individual experiments. This leads to technical batch effects which can drown out subtle biological signals. What saves us here is that each individual experiment has its own controls. (4/n)
@yuanzhi_zhu
@iScienceLuvr
Yes! Cool to see what we call the minibatch-OT trick also works in these settings!
Some works that may interest the authors (
@Yiheng_Li_Cal
@Chenfeng_X
) that also use this trick:
OT-CFM
Multisample-FM
A-OT