A Chinese research team trained a protein-focused model called ProteinTalks on tens of millions of protein measurements to forecast which single drugs and two-drug combos will kill tumor cells. Predictions worked in patient-derived cells in the lab, but clinical benefit remains untested.
ProteinTalks predicted responses in patient-derived tumor cells in vitro and found drug combinations that outperformed current approaches, but it can evaluate only two-drug combos and has not been validated in patients.
The framework transferred to other cancer types in preliminary tests, suggesting broader applicability if clinical testing and dataset expansion follow.
Triple-negative breast cancer resists many standard therapies and often requires trial-and-error treatment. The new approach builds a virtual cell focused on proteins—the molecules that perform most cellular functions—to predict what drugs a specific tumor sample is likely to respond to. If predictions hold up in patients, clinicians could choose more effective therapies earlier and avoid ineffective, toxic regimens.
Researchers at Westlake University and collaborators treated 18 immortalized breast cancer cell lines (16 triple-negative) with 63 FDA-approved anticancer drugs and 59 common drug pairs. They measured thousands of proteins at four timepoints: before treatment and at 6, 24, and 48 hours after exposure. The experiments produced more than 38 million protein measurements linked to cell survival outcomes. The dataset is open source.
Next steps for researchers and clinicians
ProteinTalks offers a proteomics-based proof of concept that a virtual cell can forecast drug responses for triple-negative breast cancer and propose effective two-drug combinations. Results in patient-derived cells are encouraging, but translating these laboratory findings into improved patient outcomes will require clinical trials and larger, more diverse datasets.
A team of researchers in China has combined machine learning, molecular docking and molecular dynamics simulations into a single computational pipeline that can sift through tens of thousands of chemical compounds and pull out a handful of genuinely promising cancer drug candidates. Writing in the journal Molecular Div
Chemotherapy has long been a workhorse against metastatic breast cancer, yet for most patients with estrogen receptor-positive (ER+) disease, the cancer eventually wins the evolutionary arms race. Cells that survive treatment pass on their resistance, and once a resistant population takes hold, the tumor becomes progre
Predicting whether a cancer patient will respond to therapy has long been one of oncology’s most consequential gambles. For people with metastatic non-small cell lung cancer, the most common and deadliest form of the disease, first-line chemoimmunotherapy can produce dramatic tumor shrinkage in some patients while leav
From: oncodaily.com For patients with hormone receptor–positive metastatic breast cancer, the therapeutic landscape has expanded dramatically. CDK4/6 inhibitors transformed first-line treatment. Molecularly selected therapies targeting ESR1, PIK3CA and the AKT pathway have created additional endocrine-based options. An
The system, designed by Stanford researchers, identified which drugs are more likely to succeed in trials and even proposed a cancer treatment a major drugmaker later landed on too. The post Virtual Biotech Company Puts 37,000 AI Agents to Work on Drug Discovery appeared first on SingularityHub .
The atlas could help scientists decipher how genetic variation shapes health and disease. The post Google’s Genome Atlas Predicts the Effect of Every Possible DNA Mutation appeared first on SingularityHub .
Loading more related stories...
Open the app view to save this story, compare related coverage, and continue from the same source.