VICTORIA, BC / ACCESSWIRE / March 8, 2024 / Anyone familiar with the intricacies of life science knows how truly complicated DNA is; without a full understanding of the language of DNA, scientists are often at a standstill when it comes to advanced medicine. That's why the work of companies like ImmunoPrecise Antibodies (NASDAQ: IPA) ("IPA") is important to identifying, preventing and treating diseases.
With over a decade of experience, IPA offers a diverse range of specialized services - including antibody discovery and development - utilizing various cutting-edge techniques and models. The company has also just announced the development of a Foundation AI Model. The model integrates the strengths of Large Language Models (LLMs) with IPA's subsidiary BioStrand's patented HYFT® Technology.
BioStrand's Foundation AI Model uses fingerprint patterns found throughout the biosphere to connect different areas of knowledge. This technology forms the backbone of BioStrand's platform, which encompasses a vast knowledge graph mapping billions of relationships across millions of data objects. By linking genetic, structural and functional data with scientific literature, BioStrand provides a comprehensive understanding of the connections between genes, proteins and biological pathways.
The integration of HYFTs with stacked LLMs allows BioStrand's AI model to decipher the language of proteins, providing valuable insights for antibody drug development and precision medicine. LLMs, originally designed for understanding natural language, can also be applied to understand the language of proteins. This enables tasks such as protein structure prediction, antibody optimization and protein mutagenesis.
To effectively analyze the language of proteins, HYFTs are used to identify meaningful units or "words" within protein sequences. This computational capability makes it possible to map and analyze these functional units accurately. These "word boundaries" represent a breakthrough in understanding protein structure and function within the language of proteins. The concept addresses a significant gap in knowledge for researchers and drug developers. By accurately identifying and manipulating functional units within proteins, this innovative approach opens doors to new possibilities in drug discovery, protein-based treatments and synthetic biology.
LLM Stacking
The unique approach called "LLM stacking" that the Advanced Foundation AI employs combines different LLMs intelligently. HYFTs are linked to specific features found in various LLMs, akin to understanding the meaning of a word based on its context. In the life sciences context, these features can include identifying amino acid residues critical for protein binding or detecting sequence variations associated with disease susceptibility. The combination of HYFTs and LLM stacking allows BioStrand to differentiate between binding and non-binding antibodies, even when they share similar HYFT patterns. The sequence diversity harnessed by HYFTs was discovered during the analysis of sequencing data sourced from Talem Therapeutics, an IPA pipeline subsidiary.
"The development of our Foundation AI Model, powered by our unique 'LLM stacking' approach and patented HYFT technology, marks a significant milestone in the field of biotechnological research," stated Dirk Van Hyfte M.D., Ph.D., Co-Founder and Head of Innovation of BioStrand. "This innovation not only expands the boundaries of current biotech research, but also establishes a new standard for the application of AI in solving complex biological challenges."
"As the global community recognizes the transformative potential of artificial intelligence in the life sciences," Dr. Hyfte continued, "I am confident that BioStrand's Foundation AI Model will stand at the forefront of innovation and the future of AI-driven solutions in biology and drug discovery."
Featured photo by Warren umoh on Unsplash.
Contact:
SOURCE: ImmunoPrecise Antibodies Ltd.
View the original press release on accesswire.com
