ESE Special topics in electrical and systems engineering: Systems Biology Pappas Kumar Rubin Julius Halász Roadmap to Systems Biology
What next? Cellular processes come down to molecular interactions –Rate laws –Kinetic constants –Differential equations … so all we need to do is get all the reactions,rate laws, constants, put them into a computer virtual cell
What next? Easier said than done: –Processes not typically known in detail –Kinetic constants Not measurable/Not measurable in vivo Meaningless –High dimensional, nonlinear systems Yet often simple behavior: emergence –Even if individual processes can be studied, the cost of going through all of them is prohibitive
What next? Biologists have “told us so”: –Reductionism doesn’t work –There are exceptions to all “laws” –Qualitative descriptions are more meaningful Source of limitations –Experimental input –Lack of fundamental understanding of processes –Lack of appropriate mathematical “language”
What next? Systems/quantitative biology today: –No mathematically expressed principles –Several qualitative principles Robustness Redundancy –Driven by experimental data –Certain clusters of modeling activity –Physics, circa 1670 (before Newton) Incremental progress on many fronts Best approach is to try to be useful to biology
Some of the fronts Genetic network identification Metabolic networks Signaling Cycles (cell, circadian) Mesoscopic / stochastic phenomena Synthetic biology Software tools
Genetic network identification Microarrays –One of the most spectacular advances in experimental technique –Typical of “high-throughput” approach –Made possible by Genome sequencing projects of the 1990’s Semiconductor, microchip technology
Genetic network identification Microarrays –Chips with a grid of RNA * microprobes –Each probe has a different sequence * –Probes represent genes –Probes hybridize to mRNA from a sample –Optical (fluorescence) readout Parallel measurement of gene expression –Commercially available for several organisms Affymetrix – “the Microsoft of biotechnology”
Gene network identification What can we learn from high throughput, semi-quantitative, perhaps time resolved, gene expression data? Identification of transcription networks –Ignore all details of interactions –Focus on the existence of an influence of Gene A onto Gene B –Various levels of abstraction, from on/off to Hill coefficients
Gene network identification Next lecture Papers by Collins, Liao A whole industry has been spawned Lots of room for new ideas coming from computer science/hybrid systems Challenge: connect with biological knowledge
Metabolic networks Another “breadth-first” approach Made possible by arduous work of many postdocs, PubMed, and other databases Metabolic reactions curated into comprehensive databases Stoichiometric information on hundreds of concurrent chemical reactions The workings of the chemical factory
Metabolic networks The state of the system is the vector of all metabolite concentrations c. Each reaction is represented by an integer vector: A + B 3X [-1, -1, 3, 0] 2A + B Y [-2, -1, 0, 1] Stoichiometric matrix S Vector of reaction rates v External fluxes of metabolites f
Metabolic networks At steady state, c is constant The state of the metabolic network is v Many possible solutions –Feasiblity cone –Which state is picked by nature? –Determined by unknown kinetic details Models postulate optimization principles
Metabolic networks Many papers: –Palsson, Church Lecture by Marcin Imielinski (?) Lots of linear algebra
Signaling Multi-cellular organisms are similar to highly organized societies –Every cell has the same genetic information –Yet they are highly specialized/differentiated –Widely different phenotypes, functions –The organism works because each cell does what it is supposed to Signaling ensures that cells act properly
Signaling In cancer, the signaling machinery breaks down –Wrong signals and/or wrong interpretation –Cells differentiate into the wrong type –They grow when they are not supposed to –Stop listening to the system commands –Take a life of their own (tumors)
Signaling Signaling tells cells to do everything –Lack of certain signals triggers cell suicide (apoptosis) Signals are carried by special molecules in the organism –Hormones, growth factors There are specialized receptors on the cell surface Receptors transduce signals (binding of their ligand) into the cytosol (the inside of the cell) Signaling cascades originate in the initial binding event Complicated networks of multistep phosphorylation reactions Eventually they control gene expression
Signaling Signaling malfunctions result from small mutations –Lack of signaling –Uninduced signals –Over/under- amplification A few well studied networks –EGF Erb/Her A few well studied cell lines
Cell signaling Huge literature Lecture: Avi Ghosh (Drexel)
Mesoscopic phenomena Face the reality of small molecule numbers Stochastic nature of reactions Well established simulation methods Often ignored, wrongly
Mesoscopic phenomena A few important results –Lambda phage (Arkin) –Lac system (van Oudenaarden) –Competence (Elowitz) Relevant experimental results –Well delimited, controlled, yet live system Lecture by Mustafa Khammash
Cycles Complicated control systems Make sure that actions are taken in the correct sequence Cell cycle –Papers by Tyson Circadian cycle –Papers by Doyle
Synthetic biology From simple genetic switches To tumor killing bacteria In between: synthesis of artemisin (Keasling)
Software Large industry Lots of potential for new work Largely ten years behind in modeling Focus on languages standardization,.. Still very important