Dami Gupta

🔬 Built IndicationScout — an agentic AI system for drug repurposing. Give it a drug name. AI agents pull live evidence from Open Targets, ClinicalTrials.gov, PubMed, ChEMBL & openFDA, then report the state of each hypothesis: live, stalled, niche, or dead. (1/2)

Science is entering a new era - one where AI agents can do scientific work. 🧬 Today NVIDIA is launching the BioNeMo Agent Toolkit - an open, agent-ready toolkit that gives any AI agent callable tools for protein structure prediction, molecular docking, generative chemistry, genomic analysis, and more. (1/2)

Full text: science.org/doi/10.1126/sc…

The hard part wasn't the APIs — it was correctness. A drug is often already approved for a narrow slice of a broad candidate disease. An approval-aware layer catches that upstream, so only genuine repurposing signal survives. Better to miss a candidate than fabricate one. Python · LangGraph · Claude API (2/2)

🔬 Built IndicationScout — an agentic AI system for drug repurposing. Give it a drug name. AI agents pull live evidence from Open Targets, ClinicalTrials.gov, PubMed, ChEMBL & openFDA, then report the state of each hypothesis: live, stalled, niche, or dead. (1/2)

Covers full academic research pipeline for Claude Code github.com/Imbad0202/acad…

@Nature @NatureBiotech Oh no … RIP

Geneticist J. Craig Venter, best known for his role in sequencing the human genome, has died aged 79. He spoke to Nature in 2023 about AI, sequencing the ocean – and why he had no plans to stop working. go.nature.com/4tHEf9M

Have you ever wondered what animal superpowers can teach us about our own DNA? Join us in Cambridge, MA (or online) for a Broad Discovery Series talk on Wednesday, May 20 at 6 pm ET with @eenork of @darwinsarkfnd. Register: broadinstitute.swoogo.com/Karlsson

Very good thread, explainer on the new LLM paper @ScienceMagazine

Anthropic just published best practices for Opus 4.7 with Claude Code. The biggest changes: Default effort is now "xhigh" (new level between high and max). Fixed thinking budgets are gone. The model decides when to think deeply on its own.

Updated: Ask a frontier genomic foundation model (Evo2) about your DNA variant ! AskEvo2 : huggingface.co/spaces/damigup… Get Evo 2 DNA variant log-likelihoods in your browser - no Docker, no GPU, no API keys, no login. Paste ref and alt sequences : get wild-type + mutant log-likelihoods. Single variant and batch modes available. Validated against ClinVar: BRCA1 AUC=0.90, MLH1 AUC=0.95, n=200 each. Results are raw Evo2 model outputs (no supervised linear probing or retraining applied). Runs via a Modal backend, hosted on Hugging Face Spaces. @goodfire_ai @arcinstitute @pdhsu @garykbrixi @NVIDIAHealth

@LukeParkerDev @plainionist The tests are ‘is not none’ and ‘is string’ !!!

@plainionist Models write the worst tests know to humanity

im am this close to crashing out. every AI just does dumb stuff unless you are so specific you may as well code. it can help for mass migrations once you've already done the shape and exact impl, and have a bunch of boring work. im so sick of trying to wrangle it lol

@Aaronontheweb Spot on !

One of the most insidious tics LLMs have when coding is this obsession with adding "fallback" behaviors everywhere These are extremely toxic because they hide real bugs and most importantly, introduce lots of potential privilege escalation vulnerabilities everywhere

@dioexul @rezoundous True dat

@rezoundous I understand the code that’s why I see vibecoding as a potential problem.

Do you understand the code or just trust the result? Be honest.

@CodveAi @unclebobmartin Or drunk interns. Or golden retriever, eager to please puppies

@unclebobmartin the trick is treating them like juniors who need context and constraints, not like seniors who'll just know what you want. structure your AGENTS.md right and they'll follow the rules.

Bottom line: For AIs all rules are more like guidelines.

@hiarun02 Well said

The real danger of AI isn’t job loss. It’s losing the habit of thinking.

@camsoft2000 Exactly, and if you express these thoughts, there will be a ton of people commenting on your lack of skills files , under using the plan mode and blah blah

I’m getting to the point with one of the projects I work on where the complexity of AI slop is becoming a real issue. While I can still happily prompt the agent to add x feature and it will do so and it will likely work perfectly, the code is just getting too complex and fragmented. Agents love to copy and paste and keeping patterns DRY is a real challenge. The agent will start diverging all those copy and pastes until you’ve got loads of similar but slightly different blocks of logic. Again it all still works and solves the problem I’m after. But I just can’t get any kind of consistency anymore, the code is a mess and I just don’t have a handle on it. I want a clean unified architecture but agents just code with tunnel vision. The project is now too big and complex for an agent to fully reason with and too big and complex for me to reason with. The only real solution is a complete rewrite. Maybe this is the way things will go. Code will just become disposable. I don’t really want to care about the code and to be honest I don’t but I do care about consistency and maintainability and the AI slop is hurting those very things I do care about. I know some will say “I’m holding it wrong”, use x,y,z skill, tool whatever and already use tools and anti slop skills, plans, docs, etc but the outcome is the same. Vibe coding something into existence is truly magical. But turning it into a mature product with months of iterations is painful. I can’t even hand code this thing because I don’t understand the code anymore and I’m too lazy to try and code myself because I’m addicted to AI. So what’s the solution, either start again and accept that’s just the way we have to roll, or just carry on fighting the slop and accept each new feature will take longer to implement than the last. I’m tired. I’m addicted.

@camsoft2000 Copy pastes ! Absolutely!

@LuizaJarovsky I don’t think it’s small

I'm convinced that a small percentage of the AI industry is already suffering some level of "AI psychosis." Maybe AI is inherently misaligned and unfit to serve any human goals.