From Generative AI to Human Clinical Trials: A story of Insilico Medicine's TNIK Program

Please do read the original paper here:
Ren, et al. A small-molecule TNIK inhibitor targets fibrosis in preclinical and clinical models. Nat Biotechnol (2024).
doi.org/10.1038/s41587-024-02...
www.nature.com/articles/s4158...
This is our first take on a documentary video explaining the story behind our novel antifibrotic program where target discovery and disease modeling AI (Biology42: PandaOmics, pubs.acs.org/doi/10.1021/acs...., was used to discover and prioritize a novel protein target for idiopathic pulmonary fibrosis(IPF), Kidney Fibrosis, Skin Fibrosis, and other diseases. Then, generative AI platform (Chemistry42: Generative Chemistry, pubs.acs.org/doi/full/10.1021...) was used to generate novel molecular structures with the desired properties. The target was also assessed using the Hallmarks of Aging Assessment (www.aging-us.com/article/2039....
Then, multiple preclinical experiments were performed to prove that the generation conditions are confirmed. Three preclinical candidates were nominated for this target and it also demonstrated activity in skin fibrosis.
Following multiple preclinical successes and IND-enabling studies, Insilico conducted a small "Phase 0" study in healthy human volunteers. Afterward, it conducted Phase I studies in New Zealand and in China and entered into two parallel Phase IIa human clinical trials in the US and in China. The complete story from zero to phase I complete is published in the Nature Biotechnology paper.
The probability of getting to this stage was very low and we decided to start shooting a documentary video to explain this story to the general public since very few companies ever attempted the complete end-to-end cycle. Many biotechnology companies either find new disease hypothesis and in-license existing drugs that someone else developed; or pick very established clinically-proven targets and develop new chemistry or antibodies from scratch.
Many companies that start this process from scratch fail without leaving significant documentary evidence and we wanted to make a substantial contribution even if the chances are not in our favor and help prepare our scientists and other groups for future attempts. But, so far, so good.
Please do enjoy the video and read the paper.
To make it easier to read this paper, we developed a conversational assistant where you can ask your questions about this paper and other papers by Insilico. We submitted the paper in June 2023, and made this assistant available for reviewers to help find relevant sections of the paper. You can see primary source with the relevant paragraphs to your right. You can now "chat with the paper" here: papers.insilicogpt.com
In addition, we also made a substantial data repository with the raw data from multiple experiments available here: insilico.com/repository/nbt-i...
Please enjoy the video and comment below.

[@bigal0172]

"Importantly, this success showed a massive milestone for our team and the broader generative AI field: from novel target discovery to Phase 1 testing in under 30 months. Relative to traditional drug discovery methodology, this AI-powered approach accomplished this feat in half the time and for a fraction of the cost." Amazing work! A true testament of the power of Generative AI.

[@RajatThomas]

Super exciting! Keep up the great work.

[@devyanipawar287]

Excellent !

[@davidmitchell3881]

Fabulous case.
Two thoughts. First could this be metastatic? Should he have a CT TAP?
Second given that the inguinal lymph nodes are moderately enlarged perhaps an ultrasound might be useful. Ultrasound is better at identifing metastatic disease than CT for superficial lymph nodes. Biopsy might also be useful as the main lesion is tricky to biopsy

[@EYE-EM-Actor-TNT-EMT-BMT]

Welcome to make branch in Bangladesh 🇧🇩