Leash Bio pushes AI molecular reasoning with experimental feedback
What happened
In a Latent Space interview, Leash Bio's Ian Quigley described how the team uses DNA-encoded libraries to collect clean, reproducible protein-small molecule interaction data and feeds it into AI scientific reasoning. Their agent-based experiment environment lets an LLM reason about structure-activity relationships in natural language, searching 6.5 million molecules for the best binder within a budget of 100 tests. Existing data supports tens of thousands of simulations with experimental feedback. Two months after recording, the environment became Minos. On October 15 the team released a preprint, blog and open dataset.
Written by AI from the coverage · updated 3 hours ago
Coverage
Follow the reports to see the story from different sides.
- Latent SpaceLab-in-the-Loop at Compute Speed — Ian Quigley, Leash Bio
Leash Bio 的 Ian Quigley 在访谈中解释了如何用 DNA 编码文库采集干净、可重复的蛋白质与小分子相互作用数据，并将其用于 AI 科学推理。团队的智能体实验环境让 LLM 用自然语言推理构效关系，在 100 次预算内从 6.5M 个分子中寻找最佳结合分子，已有数据可提供数万次带实验反馈的模拟。录制两个月后，该环境发展为 Minos，团队今天发布了预印本、博客和开放数据集。
Heat over time
Heat now 9·Comparable peak 10（Oct 15 06:00）·Comparable change over 24 hours –
The trend only compares accounts observed without gaps, so its range may be smaller than the current heat. Hover or tap the chart for each hour; the left and right arrow keys step through it.