What is genetic epidemiology and how does it establish causality?

Genetic epidemiology correlates changes in the genome with phenotypic differences. Unlike traditional observational epidemiology, it definitively establishes the direction of causality. If a gene is linked to a disease, it indicates that perturbing that gene in humans causes the disease, guiding interventions. (Timestamp: 400)

How do geneticists study disease with natural human genetic variation?

Every individual carries millions of unique genetic variants, which can be thought of as tiny perturbations. By studying large cohorts of unrelated individuals with detailed phenotypic data (including electronic health records), scientists can systematically relate millions of variants to hundreds of phenotypes. This large matrix of genotype-phenotype data is a powerful puzzle for computational biology. (Timestamp: 529)

What are the most impactful diseases to understand, considering both mortality and quality of life?

While lifestyle impacts (like mental disorders and Alzheimer's) are crucial for overall well-being, heart disease is the number one killer (650,000 Americans/year), followed by cancer (600,000). Other significant diseases include accidents, lower respiratory disease, stroke, and diabetes. COVID-19 has also become a top killer. (Timestamp: 1616)

How does understanding genetics help counteract environmental disease factors?

Genetics provides access to the underlying biological mechanisms of disease. By understanding these mechanisms, interventions (pharmaceutical or lifestyle) can be developed to directly alter the body's processes. For example, understanding how metabolism works can allow one to 'poke another hole' in the 'bathtub of fat' in obesity, bypassing simple diet and exercise. (Timestamp: 1973)

What unexpected connections has genetics revealed for diseases like Age-related Macular Degeneration (AMD) and Schizophrenia?

For AMD, genetics unexpectedly implicated the complement pathway, an immune function, suggesting an immune basis for this eye disorder. Similarly, for schizophrenia, the complement complex and microglia (immune cells involved in synaptic pruning) were unexpectedly linked to the disease, challenging previous neuron-centric views. (Timestamp: 2501)

How did epigenomics help identify the tissue of action for various diseases?

By mapping active gene regulatory elements across 127 different human tissues (Roadmap Epigenomics Consortium), researchers could correlate genetic variants with their activity in specific tissues. This revealed enrichments: type 1 diabetes variants in T/B cells, type 2 diabetes in pancreatic islets, blood pressure in the left ventricle of the heart, and surprisingly, Alzheimer's variants in microglia, not neurons. (Timestamp: 2753)

Why is it important that genetic variants often 'converge' on a small number of biological processes?

Convergence means that multiple distinct genetic lesions (perturbations) from different parts of the genome can all impact the same underlying biological process. This is a 'saving grace' for the field, as it suggests commonality and allows researchers to gain insights into fundamental disease mechanisms, even with diverse genetic origins. (Timestamp: 4210)

What are the key steps in the pipeline to go from a genetic association to a disease treatment?

The path involves six steps: 1) identifying the tissue of action through epigenomic enrichment, 2) pinpointing the causal nucleotide(s) among many co-inherited variants, 3) finding the disrupted gene regulatory motif, 4) identifying the upstream regulator that binds the motif, 5) determining the target genes affected, and 6) understanding the biological processes those genes mediate. (Timestamp: 4941)

How did researchers confirm the FTO locus's true role in obesity?

The strongest genetic association with obesity was in the FTO gene, but experiments revealed the causal variant in FTO's intron actually controlled two distant genes, IRX3 and IRX5, located 1.2 million nucleotides away. These genes proved to be master regulators of thermogenesis versus lipogenesis, dictating whether fat cells store or burn calories, ultimately explaining the obesity link. (Timestamp: 4036, 5837)

What is CRISPR-Cas9, and why is it considered revolutionary for genome editing?

CRISPR-Cas9 is a genome editing system adapted from a bacterial immune defense, which uses an RNA guide to precisely cut DNA at a specific location. Its revolutionary aspect is the ease of designing the guide RNA (simple typing) compared to previous protein-based methods (zinc fingers, TALENs), making targeted genome interventions much more accessible. (Timestamp: 5295, 5691)

What advanced technologies are driving the next phase of genomic research for disease?

Key technologies include: 1) Robotics and automation for high-throughput screening of molecules and genetic perturbations. 2) Massively Parallel Reporter Assays (MPRA, Sharper MPRA, HYDRA) for simultaneously testing thousands to millions of genetic elements. 3) Single-cell RNA and epigenome sequencing, which provide cell-type-specific resolution for gene expression and regulatory activity. (Timestamp: 6572, 6664, 7136)

How do scientists use 'barcodes' and 'bottles' for massively parallel single-cell sequencing?

Cells are dissociated and randomly distributed into 100 bottles. A unique barcode (out of 100) is added to cells in each bottle. This process is repeated three times with different randomizations. A cell's unique path through three sets of 100 bottles creates a million unique barcode combinations, labelling each cell probabilistically without microfluidics. (Timestamp: 7291)

What is 'mosaicism' in the human brain, and why is it important for understanding disease?

Mosaicism refers to genetic variations that occur after the zygote stage, leading to different cells in an individual's body (like the brain) having unique genetic underpinnings. Studying these somatic variants at a single-cell level can reveal new disease pathways, such as altered neuronal energetics in Alzheimer's, that are not visible through common genetic variations. (Timestamp: 8514)

What is 'systems medicine,' and how will it transform disease intervention?

Systems medicine moves beyond single-gene interventions to target entire networks of pathways and genes with multi-input perturbations. It combines AI for drug design, precise molecular biology tools, personalized predictions from electronic health records and polygenic risk scores, and cell-type-specific drug delivery to bring individuals from a disease state to a physiologically normal or even improved state. (Timestamp: 9081)

Key Moments

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

Lex Fridman

Science & Technology3 min read155 min video

Oct 25, 2020|251,098 views|4,036|380

manolis kellis artificial intelligence agi ai ai podcast artificial intelligence podcast lex fridman lex podcast lex mit lex ai lex jre mit ai

Save to Pod

Key Moments

TL;DR

Manolis Kellis on using human genetics to understand and combat complex diseases.

Key Insights

Human genetics is now driving basic biology, with human genome variations providing more data than model organisms.

Understanding disease requires dissecting complexity, from genetic variants to cellular functions and organismal phenotypes.

Epigenomics and single-cell sequencing are crucial tools for mapping gene regulatory elements and cell-specific functions.

Unexpected connections between diseases (e.g., immune system and Alzheimer's, complement pathway in AMD and schizophrenia) are revealed through genetics.

The path to therapeutics involves identifying causal variants, understanding regulatory circuitry, and intervening at pathway or network levels.

Future of medicine is systems-level, personalized interventions targeting complex pathways rather than single genes.

THE EVOLVING LANDSCAPE OF BIOLOGICAL RESEARCH

The traditional approach to understanding human disease relied on extrapolating knowledge from model organisms and basic biology. However, advancements in human genetics have fundamentally shifted this paradigm. Manolis Kellis emphasizes that human genetic variation now serves as a primary driver for biological discovery, offering an unparalleled data set. This genomic information allows scientists to identify perturbations and their downstream effects, moving biology from a model-organism-centric approach to a human-centric one.

DECONSTRUCTING DISEASE COMPLEXITY

Understanding human disease is an immense challenge, akin to understanding the complexity of the human genome or brain. Kellis explains that disease isn't a simple one-to-one correlation but a result of intricate interactions across multiple biological layers. The process involves mapping genetic variations to molecular changes, cellular functions, and ultimately, organismal phenotypes. This requires a systematic approach to dissecting these layers, moving from broad genetic associations to specific molecular mechanisms.

THE POWER OF EPIGENOMICS AND SINGLE-CELL ANALYSIS

Modern research tools are essential for deciphering biological complexity. Epigenomics, which studies gene regulatory elements, and single-cell sequencing, which allows for the analysis of individual cells, are transformative. These technologies enable researchers to map gene activity across vast numbers of cells and tissues, revealing how genetic variants impact specific cell types and pathways. This granular view is critical for understanding diseases where effects might be subtle or confined to particular cellular populations.

UNEXPECTED CONNECTIONS AND DISEASE MECHANISMS

Genetic studies have unveiled surprising links between seemingly disparate diseases and biological systems. For instance, the immune system's complement pathway has been implicated in age-related macular degeneration (AMD) and schizophrenia, an unexpected finding. Similarly, Alzheimer's disease, initially thought to be primarily neuronal in origin, shows significant enrichment in microglial cells (immune cells in the brain). These discoveries highlight the interconnectedness of biological systems and open new avenues for therapeutic intervention.

THE PATHWAY TO THERAPEUTICS: FROM GENETICS TO INTERVENTION

Translating genetic discoveries into treatments is a multi-step process. It begins with identifying the relevant tissue and causal genetic variants. Researchers then delve into understanding the regulatory circuitry, including specific motifs and upstream regulators, that are affected. This knowledge allows for the identification of target genes and biological pathways. The ultimate goal is to develop interventions — whether pharmaceutical or otherwise — that can precisely modulate these pathways, ideally at a systems level rather than targeting individual genes.

THE FUTURE: SYSTEMS MEDICINE AND PERSONALIZED INTERVENTIONS

The future of medicine lies in 'systems medicine,' where interventions are personalized and target complex biological networks rather than single genes. This approach leverages a confluence of technologies: artificial intelligence for drug design, advanced molecular biology for manipulation, and computational power for analyzing massive datasets. Personalized diagnostics and prognostics, informed by genetic data (common, rare, and somatic variants), will enable interventions precisely tuned to an individual's unique biological state. This promises a transformative era in disease alleviation and the enhancement of human health.

Mentioned in This Episode

●Products

●Software & Apps

●Companies

●Organizations

●Books

●Concepts

●People Referenced

Common Questions

Traditionally, biology started with model organisms (mouse, fly, yeast) to understand gene function and then applied that to human disease. However, human genetics has transformed this, with the vast amount of genetic mutation information in the human genome now driving basic biology and the understanding of disease mechanisms directly in humans. (Timestamp: 280)

Topics

Health & Longevity Neuroscience & the Brain AI & Machine Learning Drug Discovery Precision Medicine Aging Research Genome Sequencing Computational Biology Disease Mechanisms Genetic Variation CRISPR Technology Single-cell Analysis

Mentioned in this video

Software & Apps

Semrush

An advanced SEO optimization tool discussed as a sponsor.

TALENs

Transcription Activator-like Effector Nucleases, another protein-based genome editing technology considered difficult to design compared to CRISPR.

Microsoft Word

Software mentioned as an example of a successful product because its developers actively used it.

Zinc Finger Enzymes

An earlier genome editing technology that required complex engineering to target DNA sequences, predating CRISPR.

Epimap

The next generation of the epigenomics roadmap, mapping 833 different tissues and finding enrichments in 540 different disorders.

Sharper MPRA

A modification of MPRA technology that provides higher resolution by tiling elements to precisely locate regulatory control regions.

Apple Podcasts

A platform where the podcast is available for review and subscription.

Single-cell RNA Sequencing

A technology for measuring gene expression at the resolution of individual cells, invaluable for understanding cell-type-specific effects of genetic variants.

Pessimist Archive

A history podcast mentioned as a sponsor, focusing on human resistance to new things.

Microsoft Excel

Software mentioned as an example of a successful product because its developers actively used it ('eating your own dog food').

Hydra

Another modification of MPRA technology that allows testing of seven million elements simultaneously by cutting them directly from DNA, enabling broader genome accessibility analysis.

Microsoft PowerPoint

Software mentioned as an example of a successful product because its developers actively used it.

MPRA

A technology that allows simultaneous testing of thousands of genetic variants (e.g., enhancers) in parallel to determine their impact on gene expression.

Starseq

A technology developed by Alex Stark, used to couple enhancers directly with the reporter gene, simplifying readout and enabling precise measurement of enhancer activity.

condition

schizophrenia

A neurological disorder thought to involve changes in synaptic pruning, with genetic studies unexpectedly implicating the complement system and microglia.

Books

Melina Claussnitzer paper

A paper published by the collaborative team, led by Melina Claussnitzer and including Manolis Kellis, that showed the FTO locus did not implicate the FTO gene itself but rather distant genes IRX3 and IRX5, solving a long-standing obesity puzzle.

Ernst Nature paper

A paper that further expanded on the findings of epigenomic enrichment in disease-associated variants, building on previous work.

Ernst Nature Biotech paper

A foundational paper co-authored by Manolis Kellis, showing for the first time that specific chromatin and epigenomic states (enhancers) are enriched in disease-associated gene variants.

Concepts

Vienna

City where Alex Stark, former postdoc of Manolis Kellis, is currently a principal investigator.

age-related macular degeneration

A form of blindness causing progressive loss of central vision, which genetics unexpectedly linked to the complement pathway (immune function).

PRDM9 protein

A protein that guides double-stranded breaks and leads to crossovers during meiosis, influencing how genetic variations are inherited in blocks.

ARID5B protein

An AT-rich interacting domain 5B protein, identified as the upstream regulator that normally binds and represses the super-enhancer in the FTO locus, affecting IRX3 and IRX5 expression.

Cas9 protein

The protein component of the CRISPR-Cas9 system responsible for cutting DNA, which can be modified into 'dead Cas9' (dCas9) to activate or repress genes without cutting.

rs1421085 allele

A specific nucleotide variant identified as the causal variant for obesity within the FTO locus, which disrupts the binding of a repressor protein and affects thermogenesis.

Dead Cas9

A modified version of the Cas9 protein where the cutting function is deactivated, allowing it to be coupled with activators or repressors to control gene expression without making DNA breaks.

IRX5 gene

A gene likewise distantly controlled by the FTO locus variant and a master regulator of thermogenesis versus lipogenesis.

FTO gene

The 'fat and obesity-associated' gene, initially thought to be directly responsible for obesity based on strong genetic association, but later research revealed its actual mechanism involved distant genes.

IRX3 gene

A gene located 1.2 million nucleotides away from the FTO locus, found to be controlled by the regulatory variant within FTO and a master regulator of thermogenesis versus lipogenesis.

CRISPR-Cas9

A genome editing technology derived from a bacterial immune defense system, used to cut DNA at specific loci, enabling precise genome modifications for research and therapeutic purposes.

Prime Editing

A newer generation of CRISPR technology that makes single-stranded DNA breaks instead of double-stranded, allowing for more precise local repair without complete breakage.

Companies

Google

Mentioned in the context of SEO optimization, potentially alluding to search engine presence.

YouTube

A platform where the podcast is available for subscription.

McDonald's

Used as an example of an establishment offering cheap, high-calorie processed food that negatively impacts health.

Eight Sleep

A mattress that cools itself, measures heart rate variability, and has an app, mentioned as a sponsor for improving sleep.

Twitter

A social media platform where host Lex Fridman can be connected with.

Spotify

A platform where the podcast is available to follow.

BetterHelp

An online therapy service mentioned as a sponsor for professional mental health support.

NVIDIA

A technology company, lightheartedly mentioned in the context of improving graphics for a 'simulation' of reality.

Patreon

A platform where listeners can support the podcast.

Organizations

Roadmap Epigenomics Consortium

A project that built the largest map of the human epigenome across 127 tissues, revealing active gene regulatory elements and their links to disease.

Computational Biology Group

A research group at MIT led by Manolis Kellis, focused on understanding the molecular basis of human disease using computational biology.

MIT

The Massachusetts Institute of Technology, where Manolis Kellis is a professor and leads the Computational Biology Group.

Goodyear Lab

A collaborating laboratory engaged in research on multi-tissue perturbations in the context of exercise and nutritional interventions, using both mouse and human models.

People

Feng Zhang

A molecular biologist and neuroscientist, recognized for his foundational work in co-opting the CRISPR system for genome editing.

Mary Flanders

A researcher co-credited for collaboration in developing massively parallel reporter assays (MPRA) technology.

Alex Stark

Former postdoc of Manolis Kellis, now a principal investigator in Vienna, who developed the Starseek technology.

David Goggins

A public figure known for extreme physical challenges, contrasted with seeking professional therapy.

George Church

A geneticist mentioned for his humorous term 'genome vandalism' in reference to CRISPR's cutting mechanism.

Tarun Michelson

A researcher co-credited for collaboration in developing massively parallel reporter assays (MPRA) technology.

Jason Dura

A researcher whose group has contributed significantly to the development of massively parallel reporter assays (MPRA) technology.

Bob Langer

A 'giant in the field' of drug delivery systems, mentioned in the context of developing methods for delivering specific drugs to specific cell types.

Emmanuelle Charpentier

A microbiologist and Nobel laureate, recognized for her foundational work in co-opting the CRISPR system for genome editing.

Ron Weiss

A collaborator of Manolis Kellis's group who is engineering synthetic biology circuits for cancer therapeutics, using microRNA sensors to target specific cell types.

Jennifer Doudna

A biochemist and Nobel laureate, recognized for her foundational work in co-opting the CRISPR system for genome editing.

Haruki Murakami

A writer whose philosophical words on the subservient relationship between humans and genes are quoted at the end of the podcast.

Manolis Kellis

Professor at MIT and head of the MIT Computational Biology Group, focusing on the molecular basis of human disease through computational and genomic approaches.

Melina Claussnitzer

A researcher who led a team in collaboration with Manolis Kellis's group that studied the FTO locus and uncovered its true regulatory mechanism in obesity.