Pharma.ai Webinar
Epic Year-End Recap
& Q4 Winter Updates
Wednesday, December 10
10:00 - 11:00 am EST
Multimodal Multi-Omics, Multi-Species, Multi-Tissue Transformer-based Model
Aging Research, Disease Modelling, Synthetic Data Generation,
Drug Discovery, and Other Tasks
Drug Discovery, and Other Tasks
PreciousGPT is Insilico's lineup of AI models aimed at enabling digital -omics experiments
By training large language models on omics-level data, we have created a new way to run pre-clinical research at minimal costs.
Precious models empower scientists to swiftly test their hypotheses using realistic synthesized data.
Precious models empower scientists to swiftly test their hypotheses using realistic synthesized data.
- Precious2GPTa compound model combining a transformer and a diffusion architectures with omic data generation capabilities.
Read the Precious2GPT Paper → - Precious3GPTa genuinely multimodal transformer-based model trained to emulate the workflow of case-control studies, with an emphasis on chemical perturbations. We tokenize multiple data types and train one global model to perform multiple tasks in drug discovery, aging research, and synthetic data generation.
Read the Precious3GPT Paper →
Precious3GPT is our most recent and most advanced Large Language of Life Model (LLLM/LLoLM)
This unique multi-modal AI is capable of synthesizing and interpreting omics-level data from different cell lines and tissues. It combines proteomes, RNA sequencing, and DNA methylation data of multiple species under one hood and supports a variety of experimental settings. Thanks to a ChatGPT-compatible API, Precious3GPT may be integrated into custom AI-based workflows to support complex research pipelines.
Watch the demo video to see Precious3GPT in action
Originally envisioned as a tool for any-omics age prediction and conditional multi-omics synthetic data generation and "one clock to rule them all", it evolved into a comprehensive multi-modal drug discovery platform incorporating massive amounts of data from multiple areas of biology, chemistry, drug discovery. Precious1GPT model was first presented in 2022 and published in 2023 demonstrating the ability to discover protein targets with the case study on the role of Apelin Receptor in multiple age-related diseases.
Precious2GPT demonstrated the ability of high-quality data generation and is in peer-review process.Precious3GPT is a truly multi-modal platform and we are collaborating with Vadim Gladishev's lab at Harvard to validate and turn it into a community resource.
We plan to collaborate on Precious3GPT with multiple groups and then release it to the community.
We plan to collaborate on Precious3GPT with multiple groups and then release it to the community.
Precious models enable digital compound screening in human and model organism tissues
We trained our models with data on perturbations induced by 2,500 compounds in hundreds of cell lines.
By combining Precious-3 with Nach01, we can encode and represent any compound with a SMILES-defined structure and make the potential screening library virtually limitless.
By combining Precious-3 with Nach01, we can encode and represent any compound with a SMILES-defined structure and make the potential screening library virtually limitless.
Publications
Please Note: If you are seeking information about participating in a clinical trial or treatment options, please consult your healthcare provider or primary care physician.