Best Place To BuildHow Computational Microbiology drives disease research & treatment | Prof Karthik Raman | BP2B S2 E8
Karthik Raman on computational systems microbiology: modeling microbes to understand disease and health.
In this episode of Best Place To Build, featuring Karthik Raman, How Computational Microbiology drives disease research & treatment | Prof Karthik Raman | BP2B S2 E8 explores computational systems microbiology: modeling microbes to understand disease and health Systems biology uses simplified mathematical models to understand and manipulate complex biological systems, much like engineering models, while acknowledging that all models are imperfect but still useful.
At a glance
WHAT IT’S REALLY ABOUT
Computational systems microbiology: modeling microbes to understand disease and health
- Systems biology uses simplified mathematical models to understand and manipulate complex biological systems, much like engineering models, while acknowledging that all models are imperfect but still useful.
- Computational systems biology combines large-scale experimental “omics” measurements (genomics, transcriptomics, proteomics, metabolomics) with algorithms to connect molecular parts into predictive system-level behavior.
- Systems microbiology applies these ideas to microbes and microbiomes—ranging from single pathogens like tuberculosis to communities in the gut, cities, oceans, and the International Space Station—to study interactions and functions.
- Network representations (graphs) unify many problems in the lab, from metabolic pathways and microbe–microbe interactions to finding “best routes” for producing commercially valuable molecules.
- The discussion critiques hype in gut-supplement marketing and argues for more systematic, model-driven design of probiotics and microbiome interventions, including future goals like personalized or environment-specific “concoctions.”
IDEAS WORTH REMEMBERING
5 ideasModeling is selective simplification, not perfect replication.
Raman emphasizes that models mimic only key characteristics needed to answer a question (e.g., hospital beds for COVID), starting simple and adding complexity as mismatches appear.
Start with “spherical cows,” then iterate toward reality.
He defends crude assumptions as a necessary entry point, then recommends incremental refinement through the DBTL cycle when data reveal missing mechanisms (e.g., asymptomatic compartments beyond SIR).
Omics produces parts lists; systems biology builds the ‘engine.’
Genomics/transcriptomics/proteomics/metabolomics catalog what’s present, but computation is needed to infer interactions, control, and emergent behavior from those components.
Networks are the common language across scales in biology.
From microbe–microbe ecology to metabolite conversions and even atom-bond graphs of chemicals, graph abstractions enable reuse of algorithms (pathfinding, interaction inference, community design).
Drug and pathogen targeting requires a systems perspective.
Using TB as an example, choosing targets among ~4,000 genes demands understanding essentiality and off-target effects, including impacts on host proteins and beneficial microbes.
WORDS WORTH SAVING
5 quotesAll models are wrong, but some are useful.
— Prof. Karthik Raman (citing statistician George Box)
You wanna start with a spherical cow in a vacuum.
— Prof. Karthik Raman
Can we tame biology as an engineering discipline?
— Prof. Karthik Raman (citing Tom Knight)
Biology is too important to be left to the biologists.
— Prof. Karthik Raman (attributed to a major science-agency leader; described as controversial)
It’s like a forest fire… your gut nicely recovers after an antibiotic administration.
— Prof. Karthik Raman
QUESTIONS ANSWERED IN THIS EPISODE
5 questionsWhen you decide the ‘right’ abstraction level (SIR vs adding asymptomatic compartments vs digital-twin-like detail), what criteria do you use to stop adding complexity?
Systems biology uses simplified mathematical models to understand and manipulate complex biological systems, much like engineering models, while acknowledging that all models are imperfect but still useful.
In TB target discovery, what does your modeling pipeline look like end-to-end—from genome annotation to predicting essential reactions and filtering host/off-target risks?
Computational systems biology combines large-scale experimental “omics” measurements (genomics, transcriptomics, proteomics, metabolomics) with algorithms to connect molecular parts into predictive system-level behavior.
You mention functional similarity despite organismal differences in guts; how do you quantify ‘function’ computationally (pathways, metabolite fluxes, gene families), and what data are most limiting?
Systems microbiology applies these ideas to microbes and microbiomes—ranging from single pathogens like tuberculosis to communities in the gut, cities, oceans, and the International Space Station—to study interactions and functions.
What are the strongest scientific reasons to be skeptical of many consumer probiotic claims, and what evidence would convince you a product is genuinely effective?
Network representations (graphs) unify many problems in the lab, from metabolic pathways and microbe–microbe interactions to finding “best routes” for producing commercially valuable molecules.
Your lab designs ‘minimal microbial communities’ for a function—what constraints matter most (stability, invasion resistance, growth-rate matching, metabolite exchange)?
The discussion critiques hype in gut-supplement marketing and argues for more systematic, model-driven design of probiotics and microbiome interventions, including future goals like personalized or environment-specific “concoctions.”
Chapter Breakdown
Why systems biology exists: engineering thinking for living cells
The conversation opens by reframing biology as an engineering discipline: instead of isolated discoveries (one gene/protein at a time), the goal is a holistic understanding of how cellular components work together. Prof. Raman positions systems biology as the bridge between biology and quantitative/model-based reasoning needed to predict and manipulate complex living systems.
What “modelling” means (and why all models are approximations)
Prof. Raman explains modelling as building simplified representations of reality that preserve the features needed to answer specific questions. He highlights that models range from very simple abstractions to high-fidelity simulations, and that usefulness—not perfection—is the standard for a good model.
Pandemic math: SIR models, R₀, and iterative improvement
Using COVID as a familiar example, the discussion shows how epidemiological models quantify spread and inform resource planning (beds, oxygen, policy). Prof. Raman introduces the SIR model structure and emphasizes how models evolve when reality reveals missing factors (like asymptomatic infections).
From “spherical cows” to digital twins: the spectrum of fidelity
The episode contrasts playful, extreme simplifications (“assume a spherical cow”) with modern ambitions like digital twins that behave like the real system. The point: simple models are often necessary starting points, while digital twins represent an advanced endpoint for simulation and decision-making.
Microbiology modelling in practice: targeting TB and avoiding side effects
Shifting from populations to single cells, Prof. Raman describes how models help identify drug targets in pathogens like Mycobacterium tuberculosis. The challenge is selecting essential microbial genes/proteins while minimizing harm to human proteins and beneficial microbes.
DBTL for biology: design–build–test–learn and “debugging” models
The discussion maps the engineering DBTL cycle directly onto computational biology workflows. Model predictions are tested against data, deviations are treated as clues, and the model is iteratively improved—similar to debugging a complex engineered system.
Omics primer: genomics, transcriptomics, proteomics, metabolomics
Prof. Raman introduces omics as whole-system measurement technologies that catalog the cell at multiple layers—from DNA to RNA to proteins to small molecules. Computational systems biology then focuses on assembling these parts into coherent, predictive models of function and regulation.
Microbiomes across environments: gut, deep-sea extremophiles, and the ISS
The episode broadens from the gut microbiome to environmental microbiomes in extreme and engineered settings. Studying diverse ecosystems helps discover exotic metabolisms and pathways that can be repurposed for industrial and medical applications.
Why exotic microbes matter: green manufacturing and metabolic engineering
Prof. Raman connects environmental microbiology to sustainable industry: microbes can be engineered to manufacture drugs and chemicals. He describes metabolic engineering as rerouting cellular pathways toward a “molecule of interest,” including using co-cultures for division of labor.
Networks as the unifying language: from pathways to Google Maps for metabolism
Networks/graphs are presented as the recurring conceptual tool across the lab’s work. From microbe–microbe interactions to metabolite reaction networks, network algorithms enable pathfinding and optimization—similar to routing on Google Maps, but inside cells.
IBSE at IIT Madras: interdisciplinary center and national-scale genomics
Prof. Raman explains the origin and evolution of IBSE into a center focused on integrative biology and systems medicine. He highlights major initiatives including Genome India (sequencing/analysis) and translational projects like improved gestational-age models for Indian populations.
Metagenomics & city-scale surveillance: from subway swabs to sewage signals
The conversation explores metagenomics as sequencing the genetic material from all organisms in a place, enabling microbial “signatures” of environments. Projects like MetaSub extend to Chennai, and the discussion points to future genomic surveillance via water/sewage and even airline waste to detect outbreaks early.
Gut microbiome reality check: antibiotics, recovery, and supplement marketing
Prof. Raman reframes the gut as a major metabolic organ and explains how antibiotics can disrupt it like a “forest fire,” with recovery taking months to years. He discusses why off-the-shelf probiotic claims can be oversimplified, and why modelling is needed for more systematic, personalized interventions.
Reverse engineering life: “Can a biologist fix a radio?” and ethics
The episode contrasts engineering’s build-from-known-components approach with biology’s reverse-engineering reality. Modelling helps explore “what-if” questions when direct experimentation can be impractical or unethical, especially in higher organisms.
Career path and the future: dynamics, invasion, and designed microbial communities
Prof. Raman traces his transition from chemical/food engineering to computational science and academia, arguing that computation becomes most exciting when applied to a rich domain like biology. He closes with the lab’s forward-looking agenda: modelling microbiome dynamics, designing minimal functional communities, and creating targeted probiotic-like interventions for humans and even corals.
EVERY SPOKEN WORD
Install uListen for AI-powered chat & search across the full episode — Get Full Transcript
Get more out of YouTube videos.
High quality summaries for YouTube videos. Accurate transcripts to search & find moments. Powered by ChatGPT & Claude AI.
Add to Chrome