newsfilter.io
Interview

Manolis Kellis: Biology of Disease | Lex Fridman Podcast #133

  • Computational engineering for artificial general intelligence and consciousness may require a fundamental rethinking of current system architectures.
  • Research into disease complexity across the genome, epigenome, brain circuitry, immune system, and cancer is expected to deepen understanding of these biological systems.
  • Human genetics is predicted to surpass traditional animal models as the primary driver of basic biological research.
  • Interventions via medications or lifestyle changes are anticipated to target specific disease mechanisms identified through detailed molecular phenotypes.
  • Individuals carrying genetic perturbations on shared pathways will be a target for therapeutic assistance.
  • Genetic variation will be systematically correlated with molecular variation at expression, epigenomic, and cellular levels to analyze thousands of traits.
  • Complex diseases like Alzheimer's will be deconvoluted by breaking them into single-cell level experimental puzzles to resolve brain region and cell-type complexity.
  • Endophenotypes including brain activity, heart rate, and cognitive behaviors will be measured to trace causality and disease progression.
  • Lifestyle interventions such as altering food availability and increasing social support are expected to improve public well-being and reduce mortality rates.
  • Understanding genetic components of disease is projected to mitigate environmental factors by altering underlying biological mechanisms.
  • Pharmaceutical interventions are predicted to induce a 70% change in genomic expression to counteract weak effect mutations that evolution has only tolerated a 2% change in.
  • Genes exhibiting very little variation are expected to be developmental or early embryonic lethal due to evolutionary exclusion of mutations.
  • Coupling genetics with expression variation will identify optimal pharmaceutical intervention sites even when genetic variants slightly alter gene expression.
  • A genomicist is expected to become a specialist in every single disorder as disease studies unify through genetics.
  • A "circuitry group" is planned to solve the fundamental circuitry of the human genome required to address schizophrenia, Alzheimer's, metabolic disorders, and immune disorders.
  • 93% of genetic variants associated with disease fall in non-coding regions and do not directly impact proteins, creating difficulty in linking variants to target genes via long-range circuitry.
  • Specific enhancers and master regulators, such as the OB1 enhancer, are expected to be identified as controlling distant genes like IRX3 and IRX5 through long-range interactions.
  • The causal variant RS1421085 is expected to be edited using CRISPR-Cas9 to reverse an obese phenotype to a lean one.
  • Thermogenesis and lipid metabolism are anticipated to be identified as master processes regulated by genes acting as a "dissipation knob" for energy.
  • Research methods will shift from studying single loci to systematically testing thousands of loci using automation, robotics, and massively parallel reporter assays (MPRA).
  • Hydra technology is planned to test 7 million accessible regions of the genome simultaneously by cutting directly from DNA.
  • Single-cell RNA sequencing and epigenome profiling will enable the reconstruction of regulatory circuitry across millions of cells from thousands of individuals.
  • Massive datasets, including 10 million cells from the human brain across a dozen disorders, will be generated to identify keys to understanding schizophrenia and Alzheimer's.
  • Computer scientists with expertise in machine learning and inference are deemed necessary to decouple massive, noisy biological data matrices.
  • The timeline from discovery to therapeutics is predicted to shorten from 50 years to 10 years due to the convergence of new technologies.
  • The 21st century is expected to be remembered for the dramatic manipulation of human biology, potentially curing diseases like Alzheimer's.
  • Synthetic biology constructs utilizing microRNA sensors are expected to enable specific interventions in tumor, immune, or stromal cells.
  • Different classes and types of microglia playing distinct roles in Alzheimer's versus schizophrenia are expected to be discovered.
  • Pseudo-temporal models will be used to trace Alzheimer's progression from the entorhinal cortex to the hippocampus and neocortex to create early diagnostics and biomarkers.
  • Somatic variants occurring after the zygote are expected to reveal neuronal energetics and oligodendrocyte functions not visible in common variants.
  • Pushing human lifespan beyond 120-150 years is expected to reveal new challenges regarding cancer, Alzheimer's, and metabolic disorders.
  • The next stage of medicine is predicted to be "systems medicine," intervening at the network level to avoid on-target side effects from pleiotropic genes.
  • AI and graph neural networks will be utilized for drug design and personalized medicine prediction using polygenic risk scores weighted by common, rare, and somatic variants.