Information Processing Quick Bayes Rule Derivation Information Processing
Information Processing Quick Bayes Rule Derivation
Information Processing • Example: – Cell needs enzyme A to metabolize sugar – Enzyme A has a high cost to produce • If there is little sugar outside cell, enzyme A is “not worth it” – Cell can sense internal state, not external – Cell must guess external state from internal • Well tuned probabilistic response
Information Processing Sense!
Information Processing Sense!
Information Processing • Probability Distributions on number of sugar molecules in cell given: – Low external concentration – High External concentration • Bayesian probability of high concentration given that quantity
Information Processing • Probability Distributions on number of sugar molecules in cell given: – Low external concentration – High External concentration • Bayesian probability of high concentration given that quantity
Information Processing • Probability Distributions on number of sugar molecules in cell given: – Low external concentration – High External concentration • Bayesian probability of high concentration given that quantity – Note lowered likely hood at high quantity
Information Processing • Different (internal) sensors were modeled – Sensors with different numbers of binding states – Single, double and cooperative sensors – Repressor and activator type sensors • Results – More binding states are better – Single is worst; Cooperative is best – Repressors work a bit better than activators
Information Processing
Information Processing • Simulated two state inference in “real-time” • Shows good predictive ability • Shows repressors (blue) slightly better than activators (red)
Information Processing • Cells need to infer exterior state from interior • Have evolved good estimates – Use internal sensors – Can keep track of priors • Build up of proteins on outer membrane • Cells can track this well in real time • Also gives me ideas on implementation of my algorithm
- Slides: 11