Protein Hypernetworks Johannes K oster, Eli Zamir, Sven Rahmann - PowerPoint PPT Presentation

Genome Informatics Protein Hypernetworks Johannes K¨ oster, Eli Zamir, Sven Rahmann August 20, 2012 1 / 14

Protein Network Modelling Genome Informatics B A C G H Interaction maps (undirected graphs) F D I E 2 / 14

Protein Network Modelling Genome Informatics B A C G H Interaction maps (undirected graphs) F D I E Differential equations d[ C ] (Law of Mass Action), d t = k on [ A ][ B ] − k off [ C ] Bayesian Networks, ... 2 / 14

Protein Network Modelling Genome Informatics B A C G H Interaction maps (undirected graphs) F D I E scalability accuracy Differential equations d[ C ] (Law of Mass Action), d t = k on [ A ][ B ] − k off [ C ] Bayesian Networks, ... 2 / 14

Protein Network Modelling Genome Informatics B A C G H Interaction maps (undirected graphs) F D I E scalability accuracy ? Protein Hypernetworks Differential equations d[ C ] (Law of Mass Action), d t = k on [ A ][ B ] − k off [ C ] Bayesian Networks, ... 2 / 14

Structure Genome Informatics 1 Protein Hypernetworks 2 Mining Protein Hypernetworks 3 Data Aquisition 3 / 14

G H I A G B Idea Genome Informatics Protein Network ( P , I ) B A C G H F D I E 4 / 14

Idea Genome Informatics Protein Network ( P , I ) � B A G H C I G H F A D G I B E 4 / 14

Idea Genome Informatics Protein Hypernetwork ( P , I , C ) Boolean Logic Protein Network ( P , I ) � Constraints C B A G H { G , H } ⇒ { I , H } C I G H F A D { A , B } ⇒ ¬{ G , B } G I B { G , B } ⇒ ¬{ A , B } E 4 / 14

B A A B G G G H I Mining Protein Hypernetworks Genome Informatics G B A H C I G Protein Hypernetwork ( P , I , C ) H F D A I G B E 5 / 14

Mining Protein Hypernetworks Genome Informatics G B A H C I G Protein Hypernetwork ( P , I , C ) H F D A I G B E Minimal network states ( Nec , Imp ) for q ∈ P ∪ I A B C A A B � H q ∧ c G H F G C A D G H I G c ∈ C E G D B I F F I F E I D C Satisfying model α : P ∪ I → { 0 , 1 } by tableau algorithm Nec := { q ′ ∈ P ∪ I | α ( q ′ ) = 1 } Imp := { q ′ ∈ P ∪ I | α ( q ′ ) = 0 due to active c ∈ C } 5 / 14

Tableau Algorithm Genome Informatics Propositional Logic Tableau Algorithm for a given formula f AB ∧ ( AB ⇒ BC ) ∧ ( CD ⇒ ¬ DE ) explore depth-first the tree of deductions from root f AB each root-leaf-path without ( AB ⇒ BC ) contradiction is a satisfying model ( CD ⇒ ¬ DE ) ¬ AB � BC ¬ CD � 6 / 14

Tableau Algorithm Genome Informatics Propositional Logic Tableau Algorithm for a given formula f AB ∧ ( AB ⇒ BC ) ∧ ( CD ⇒ ¬ DE ) explore depth-first the tree of deductions from root f AB each root-leaf-path without ( AB ⇒ BC ) contradiction is a satisfying model ( CD ⇒ ¬ DE ) Modifications ¬ AB � BC expand disjunctions from ¬ CD � left to right allow to pre-block subformulas to guide the algorithm to the right model 6 / 14

G H B A G I Minimal Network States Genome Informatics Clashes Two minimal network states ( Nec , Imp ) and ( Nec ′ , Imp ′ ) are clashing iff Nec ∩ Imp ′ � = ∅ or Nec ′ ∩ Imp � = ∅ . not clashing pair → interactions simultaneously possible 7 / 14

Minimal Network States Genome Informatics Clashes Two minimal network states ( Nec , Imp ) and ( Nec ′ , Imp ′ ) are clashing iff Nec ∩ Imp ′ � = ∅ or Nec ′ ∩ Imp � = ∅ . not clashing pair → interactions simultaneously possible G H B A + = � G I 7 / 14

Minimal Network States Genome Informatics Clashes Two minimal network states ( Nec , Imp ) and ( Nec ′ , Imp ′ ) are clashing iff Nec ∩ Imp ′ � = ∅ or Nec ′ ∩ Imp � = ∅ . not clashing pair → interactions simultaneously possible B B A A + = � G G 7 / 14

A B A B C G G H H F D I I I E A C G H F D I I E Prediction of Protein Complexes Genome Informatics B A Network based complex prediction C G H ◮ e.g. dense regions F D I E 8 / 14

A C G H F D I I E Prediction of Protein Complexes Genome Informatics B A Network based complex prediction C G H ◮ e.g. dense regions F D I E A B A B C Maximal combinations of minimal G G H H F D network states I I I E 8 / 14

Prediction of Protein Complexes Genome Informatics B A Network based complex prediction C G H ◮ e.g. dense regions F D I E A B A B C Maximal combinations of minimal G G H H F D network states I I I E A C Refined complexes G H F D ◮ no violated constraints I I E 8 / 14

B A AB BC BG AG AH C G H GH CF FG CD F HI DF D FI I EF ED EI E Prediction of Functional Importance Genome Informatics Minimal network state graph Minimal network states B A AB A B C A B A BC BG AG H AH C G H F G G C H GH A D CF G FG CD H I G F HI DF D G E FI D B I F I I F EF ED F EI I E D C E 9 / 14

Prediction of Functional Importance Genome Informatics Minimal network state graph Breadth first search from each node B B A AB A AB BC BC BG BG AH AG AH AG C C G G GH H GH H CF CF CD FG CD FG F HI F HI DF DF D D FI FI I I EF EF ED ED EI EI E E 9 / 14

Prediction of Functional Importance Genome Informatics Minimal network state graph Breadth first search from each node B B A AB A AB BC BC BG BG AH AG AH AG C C G G GH H GH H CF CF CD FG CD FG F HI F HI DF DF D D FI FI I I EF EF ED ED EI EI E E Perturbation Impact Score PIS ( P , I , C ) ( Q ↓ ) := | BFS ( Q ↓ ) | 9 / 14

Harvesting Constraints Genome Informatics Text-Mining Observation: Interaction dependencies are reported as single sentence natural language statements in literature. Tokenize full-text papers into relevant words and search for simple regular expression patterns. 71 new interaction dependencies from 59 000 human adhesome related papers. K¨ oster, Zamir, Rahmann. 2012 ... binding of Abl induces a conformational change in Cbl that allows binding of Src ... p . i p a p i p d i p p p . 10 / 14

Protein Hypernetworks Johannes K oster, Eli Zamir, Sven Rahmann - PowerPoint PPT Presentation

Genome Informatics Protein Hypernetworks Johannes K oster, Eli Zamir, Sven Rahmann August 20, 2012 1 / 14 Protein Network Modelling Genome Informatics B A C G H Interaction maps (undirected graphs) F D I E 2 / 14 Protein

Protein Hypernetworks Johannes K oster TU Dortmund, Informatik LS 11 Max-Planck-Institute of

Transient sequences in adaptive spiking networks: hypernetworks and spatiotemporal processing

Protein-Protein interactions Reducing the complexity Why are protein-protein interactions

DNA RNA Protein synthesis AMINO ACIDS PROTEIN Protein degradation FUNCTION Some properties

Dynamics of Protein-Protein Interactions: A Probabilistic Model Toward Protein Function Amir

PROTEIN EXPRESSION AND PURIFICATION PROTEIN EXPRESSION AND PURIFICATION Why do we decide to

Hasup Lee, Seungtaek Sun and Ye-Yeong Park ( Group 6 ) Protein-Protein interaction is

Protein Folding Protein Folding Proteins have unique 3-dimensional shapes created by the

Protein Folding Protein Folding Proteins have unique 3-dimensional shapes created by the

N C C C protein sequence but is not fully rigid C C peptide C C bond

DNA What is the major compon onent nt of all cells? s? PROTEIN INS Why would protein

Protein Gaussian Image (PGI): a protein structural representa6on

High Throughput Selection of Stable Protein Variants Using Green Fluorescent Protein to Quantify

Challenges for Protein Processing Diversity Denis CHEREAU January the 23 rd 2019 Challenge for

Protein NMR What do you need for the assignment? To prepare isotopically labelled protein ( 13

Protein Docking and 3D Ligand-Based Virtual Screening Modeling Protein Flexibility Using Elastic

WHATS T-REX? Control the synthesis of any protein of interest Silence the protein

Collaboration-based Function Prediction in Protein-Protein Interaction networks Hossein Rahmani

Making a Claim: Factors impacting Protein Quality and a New Way for Measuring David W. Plank May

Protein Structure Prediction 1 Ram Samudrala, University of Washington Rationale for

Predicting Protein Folding Paths S.Will, 18.417, Fall 2011 Protein Folding by Robotics S.Will,

Protein threading Protein Threading Basic premise Structure is better conserved than

Identification of direct residue contacts in protein-protein interactions from multi-species

The ITI TXM Corpora: Tissue Expressions and Protein-Protein Interactions Bea Alex , Claire Grover,