protein interaction network

protein interaction network

1. function: catalyst, transport,store other mol,provide strength, confer immunity, signal, growth & dev

   • function determined by 3D str, which is determined by aa seq
     • function form: 1: with small mol: lipid etc
     • function form: 2: pro-pro polymer: structure(cytoskeleton)/transcription/splicing/translation/cell cycle control/secetion/enzyme complex
       • enzyme complex: (metabolic channeling)(static pro interaction)eg: glycolysis/fatty acid biosysthesis
       • but, also transient interaction:(hard to study) enzyme with target protein . eg: protein kinase

2. force mediate the PPI include: electrostatic interaction, hydrogen bond, van der Waals, hydrophobic effect

   • the hydrophobic force drive PPI. the H bond and salt confer specificity
     • 1: a small region fitting into a cleft of partner
     • 2: two surface interact via a large area

3. experimental method: use the data to generate PPI

   • 1: yeast two-hybrid system
     • basic principle: X pro fused to DNA binding domain (DB), Y fused to activation domain (AD), see if reporter gene experssed
     • automation, large scale two-hybrid study: genome wide: bait protein and prey protein are transferred to opposite mating type, let then mate, see if form diploid
       • 1: the matrix approach
       • 2: the library screening approach
   • 2: AP-MS
   • 3: protein array/ 4: synthetic lethality

4. however, noise and systematic biases

   • a lot false postive

5. result: topology of network, finding hub,analyse subnetwork based on common motif

   • small-work = any two vertice can be connected just by a few links
   • scale-free = a few highly connected hub + may less connected vertice( P(K)~k to the power of law)
   • predict sub-/module : 1) network decomposition= k-coredecomposition. 2) algorithm that cluster modules

6. contrast to experimental method, use computational method to predict PPI

   • assumption: protein pairs are more likely to functionally or physiclally interact when their corresponding gene locate close or their arrangement is conserved
     • domain fusion, interacting protein tend to co-evolve
       • 1: co-evolution of protein family =similarity of phylogeneric tree
       • 2 co-evolution = the accumulation of the same/correlated mutation between them
       • 3: also possible to predict just from aa seq info(then know charge,surface tension, hydrophobicity), machine learning
       • 4: extract info from literature: 1) machine learning screen the interaction info + by hand. 2) natural language processing /stattiscal analysis of co-occurance

7. DB: DIP/ BIND/ MINT