Andrew Hanneman, PhD
What’s Hot in 2021: Artificial Intelligence
Machines that can solve protein 3D structures are changing protein-based drug development
Protein based (bio)therapeutics represent an advance over traditional small molecule drugs as they are responsible for much of the routine metabolic work in our bodies (think muscle tissues, enzymes, blood clotting factors, cell surface markers, signaling molecules, etc). However, proteins are also complex large molecules composed of 20 or more unique amino acids hundreds or even thousands of amino acids in length that fold into a complex 3D structure required for activity.
Proteins, in contrast to DNA, cannot be amplified using tools like the well-known polymerase chain reaction (PCR). Rather the study of protein structure relies on brute-force analytical methodologies often involving multiple rounds of purification followed by application of a variety of bioanalytical chemistry techniques including NMR spectroscopy, X-ray crystallography, and mass spectrometry. These are all fields of expertise requiring years of training. The need for these protein analytical methods has evolved due to in large part to the critical need for understanding protein 3D structure required to design and optimize protein therapeutics.
Into the breach has recently arrived artificial intelligence (AI), not only capable of regularly beating chess grand masters, but lately to solve protein 3D structures de novo given only an amino acid sequence. This new capability recently achieved by the DeepMind algorithm will fundamentally change protein-based drug development in the coming years by shifting much of the complex time-consuming bench-level protein chemistry used to determine protein structure onto the backs of nimble AI tools.
Andrew Hanneman, PhD, Scientific Advisor, Biologics Testing