Our 🪐SATURN method is now out in
@naturemethods
!
SATURN paves the way for universal cell embeddings, enabling integration of datasets across different species 🐒🐁🧍🐟🐸 Using protein language models, we encode biological meaning of genes in scRNA-seq datasets.
SATURN performs cross-species integration and analysis using both single-cell gene expression and protein representations generated by protein language models.
@jure
@YanayRosen
@mariabrbic
@yusufroohani
I'll be joining EPFL
@EPFL_en
as an assistant professor this fall! I'm immensely grateful to my mentors, collaborators, colleagues and everyone who supported me. Thank you all!
Excited to continue my research at the intersection of ML and biomedicine!
Happy to share that our work has been accepted to
#ICLR2022
! We introduce a novel open-world semi-supervised learning setting and propose ORCA method that simultaneously recognizes existing and discovers novel classes. W/
@kaidicao
and
@jure
Paper:
Introducing STELLAR: a geometric deep learning method for cell type discovery and identification in spatially resolved single-cell datasets.
@jure
@GarryPNolan
@SnyderShot
What is needed for generalizable and interpretable few-shot learning methods applicable across many domains?
We advocate that the answer lies in concept-based reasoning!
Check our paper at
#ICLR2021
poster session today, 5-7pm PDT!
w/
@kaidicao
and
@jure
How to infer human labelling of a given dataset in a model-agnostic way?
Check our new method HUME accepted at
@NeurIPSConf
as
#spotlight
!🌟 HUME provides a new view to tackle unsupervised learning.
Kudos to my fantastic PhD student
@artygadetsky
!
Paper
Excited to share our PlaNet 🌏method that reasons over population variability, disease biology, and drug chemistry using a massive clinical knowledge graph. We apply PlaNet to predict outcomes of clinical trials.
w/
@michiyasunaga
@agrwalprabhat
@jure
Honored and humbled to receive the Early Career Award 🏆. Thank you
@ISBSIB
for this recognition!
Immensely grateful to my mentors and colleagues for the support through this journey🌟
For her research bridging machine learning and biomedicine and her strong involvement in promoting diversity, equity and inclusion in computer science, Maria Brbic (
@mariabrbic
) from
@EPFL_en
receives the 2023 Early Career Award! Congratulations!
@epflSV
#SIBawards
#bc2basel
Interested in how to bridge labeled and unlabeled data in biology and biomedicine? We are organizing a tutorial on Meta-learning for biomedicine at
#ISMB
. Stay tuned for slides and updates!
w/
@chelseabfinn
and
@jure
More details at:
Can we create universal cell embeddings that capture the biological meaning of genes?
We present SATURN🪐 a method that integrates datasets across species by coupling RNA expression with protein embeddings.
w/
@mariabrbic
@yusufroohani
@jure
Preprint:
Our Aging Fly Cell Atlas is out. Do different cell types age at the same rate? No. Which cell types age faster? Fat, liver and muscle. Which cell types age slower? Neurons. Great collaboration with
@StephenQuake
@HenriJasper
@mariabrbic
@TzuChiaoLu
Video of our ISMB/ECCB tutorial on Meta-learning for Bridging Labeled and Unlabeled Data in Biomedicine is released by
@iscb
! With
@jure
and
@chelseabfinn
Link:
Tutorial website with slides and additional information:
We are accepting submissions for ICML Computational Biology workshop!
#ICML2023
#Hawaii
#ComputationalBiology
Your submissions are highly welcomed! Submission deadline is May 17th, 2023.
📢 Calling all
#CompBio
enthusiasts! Our ICML 2023 Computational Biology Workshop is now open for submissions! Don't miss this opportunity to showcase your cutting-edge research in this exciting field. check more details at
#ICML2023
#ComputationalBiology
Consider submitting your work to the Machine Learning for Genomics Explorations (MLGenX) workshop at
@iclr_conf
! Join us to unite the worlds of
#MachineLearning
&
#Genomics
🧬
⏰Submission deadline: February 4, 2024 (AOE).
📢🚨 Excited to announce the Machine Learning for Genomics Explorations (MLGenX) workshop at
@iclr_conf
2024.
⏰ Submission deadline: February 4, 2024 (AOE).
🔗 Call for papers:
🗓️ Date: May 11, 2024.
Looking forward to giving invited talk at
#KDD
Workshop on Knowledge Graphs today 5:30pm ET!
I will talk about our work on constructing knowledge graph for predicting outcomes of clinical trials.
First first whole-organism single-cell atlas of the fly is released! It was a wonderful experience to be part of the team. Special shout-out to
@HongjieLi5
for his contributions on this tremendous effort!
Data portal:
Paper:
Stanford Neuro-omics initiative hosts a virtual workshop on transcriptomics and proteomics (Nov 10-12 and 17-19
@Stanford
). Liqun Luo, Steve Quake, Jure Leskovec
@jure
,
Alice Ting
@aliceyting
and Sergiu Pasca labs. Registration link:
Join us for the first of our
#AI
in Individualized Medicine Speaker Series on April 7! The webinar, featuring
@mariabrbic
& hosts
@JohnKalantari
& Kia Khezeli will focus on Meta-Learning for Novel Cell Discovery in Single-Cell Experiments. Register here:
Excited that our COMET framework for concept-based meta-learning has been accepted to
#ICLR2021
! Joint work with
@caokd8888
@jure
Paper:
Website:
Code:
Thrilled to announce that I will be joining
@MetaAI
next month as a Research Scientist 😍
I will be working in the Brain & AI team on decoding language from neural activity, to hopefully help those which have difficulties to speak or type. Learn more here:
We apply PlaNet to reason about outcomes of clinical trials by structuring the clinical trials database and incorporating it in our clinical KG. We show that PlaNet effectively reasons about drug efficacy and safety, even for experimental drugs and their combinations.
Sharing great news – Excited to announce I will join
@Duke
as an assistant professor in
@DukeBME
starting January 2024! My lab will leverage computational, biomaterial, and spatial-omics tools to probe/control tissue structure in situ, focusing on advancing cell therapies (1/8).
PlaNet defines a universal framework that can be applied to a wide range of pharmacological tasks such as predicting drug efficacy and safety, and reasoning about population characteristics that affect drug outcomes.
HUME is a model-agnostic framework for inferring human labeling of a given dataset without any external supervision. It is compatible with any large pretrained and self-supervised model and requires training only linear models on top of pretrained representations!
The key insight behind HUME is that classes defined by many human labelings are linearly separable regardless of the representation space. We use this insight to define generalization-based objective and show it is extremely well correlated with underlying human labeling.