presenting EvoDiff: new generative models for controllable protein design from sequence data alone!
✅generate high-quality proteins
✅scaffold functional motifs
🤩apply to therapeutic design + more!
👉
💻
🎙️
EvoDiff combines evolutionary-scale data with diffusion models for controllable protein sequence generation.
In addition to generating plausible proteins, we can scaffold structural motifs in sequence space!
Preprint:
Code:
Our work on developing responsive nanoparticles that detect
#LungCancer
as a urinary readout is now published in
@ScienceTM
! We use
#MachineLearning
for accurate classification of lung cancer in mice.
Paper 👉
MIT News 👉
Excited to present our work on guided drug discovery and prediction using evidential uncertainty!
@NeurIPS2020
#ML4Molecules
Spotlight talk 👉
Poster 31 @ 4pm EST today 👉
Drop by with questions or to discuss!
⭐️ our latest work in
@ScienceMagazine
🩸 early
#CancerDetection
from only a blood draw by combining liquid biopsy + molecular eng
🧬 we engineer nanoparticles and antibody priming agents to boost cfDNA for improved diagnostics
We engineer peptides to sense targeted enzyme activity in cancer tissue! Published in
@NatureComms
:
✅ nanoscale + noninvasive
✅ high-fidelity cancer monitoring
✅ multi-scale resolution sensing
from single cell 🦠 - to tissue 🫁 - to whole organism🧍
👉
Grateful to have an amazing
@MIT
news feature on our Intro to Deep Learning course!
We cover emerging topics in
#DeepLearning
including bio/med applications (this year, some of my recent work with
@snbhatia
!). Co-taught and organized with
@xanamini
.
🌟🌟
KI researchers developed enzyme-targeting nanoprobes that could prove useful in both the clinic and the lab.
The nanosensors track tumor progression and treatment response in real time, and map enzyme activity to precise locations within cancer tissue.
(1/7) In the first of two articles online with week,
@apsoleimany
@xanamini
and
@samgoldman19
illustrate the application of a recent approach to uncertainty quantification, "evidential uncertainty", to chemical tasks in
@ACSCentSci
|
AI, the defining technology of our time, is uniquely positioned to help people solve some of the world’s most challenging issues. Here’s how some of our researchers are using it to help power their work.
Our new work developing nanosensors to diagnose
#pneumonia
is now out in
@PNASNews
! We use machine learning to classify directly from molecular barcodes.
📄 Paper:
🤩 Press:
🖥️ Code:
Led by
@_melodi_a_
!
applications for internships with our BioML group at
@MSFTResearch
are due by 11:59pm PT today December 1!!
please apply with a research statement so we can read your ideas!
work w/ me,
@KevinKaichuang
@alexijielu
@lorin_crawford
and Kristen Severson
Curious about next-generation diagnostics for cancer and infectious disease? Check out our new review in
@TrendsMolecMed
on engineering activity-based diagnostics as tools for precision medicine, using proteases,
#CRISPR
,
#synbio
,
#ML
, & more! 🧬
Paper 👉
I wholeheartedly pledge to oppose racism and hate through active allyship, continuous unlearning, learning, and listening. We must work together to achieve inclusion & equity in academia and beyond. Passing to
@snbhatia
@DrTarekFadel
@jessekirk15
@cnloynachan
@BiancaLepe
I wholeheartedly pledge to oppose racism and hate through active allyship, continuous unlearning, learning, and listening. We must work together to achieve inclusion & equity in academia and beyond. Passing to
@optiML
@animesh_garg
@apsoleimany
@duckietown_coo
@ramin_m_h
We developed a fast, calibrated method for estimating the uncertainty of neural networks in key regression learning problems.
Check out the paper link or thread below for more details!
Our preprint "Deep Evidential Regression" is online!
We present a fast method for learning uncertainty of neural networks without sampling, by placing evidential priors over the likelihood and training the NN to infer hyperparameters of the prior. 1/
👉👉