Three-Stage Prediction of Protein Beta-Sheets Using Neural Networks, Alignments, and Graph Algorithms

0
0
2650 days ago, 859 views
PowerPoint PPT Presentation

Presentation Transcript

Slide 1

Three-Stage Prediction of Protein Beta-Sheets Using Neural Networks, Alignments, and Graph Algorithms Jianlin Cheng and Pierre Baldi Institute for Genomics and Bioinformatics School of Information and Computer Sciences University of California Irvine

Slide 2

Importance of Predicting Beta-Sheet Structure Ab-initio Structure Prediction Fold Recognition Model Refinement Protein Design Protein Folding Coil beta-sheet helix Rendered in Protein Explorer

Slide 3

An Example of Beta-Sheet Architecture Level 1 4 5 2 1 3 6 7 Structure of Protein 1VJG Beta Sheets

Slide 4

An Example of Beta-Sheet Architecture Level 1 Level 2 4 5 Antiparallel 2 1 3 6 7 Parallel Strand Pair Strand Alignment Pairing Direction Structure of Protein 1VJG Beta Sheets

Slide 5

An Example of Beta-Sheet Architecture Level 1 Level 2 Level 3 4 5 Antiparallel H-security 2 1 3 6 7 Parallel Strand Pair Strand Alignment Pairing Direction Structure of Protein 1VJG Beta Sheets Beta Residue Pair

Slide 6

Previous Work Statistical potential approach for strand arrangement (Hubbard, 1994; Zhu and Braun, 1999) Statistical possibilities to enhance beta-sheet auxiliary structure expectation (Asogawa,1997) Information hypothesis approach for strand arrangement (Steward and Thornton, 2000) Neural systems for beta-buildup sets (Baldi, et al., 2000)

Slide 7

Three-Stage Prediction of Beta-Sheets Stage 1 Predict beta-deposit blending probabilities utilizing 2D-Recursive Neural Networks ( 2D-RNN, Baldi and Pollastri, 2003 ) Stage 2 Use beta-buildup matching probabilities to adjust beta-strands Stage 3 Predict beta-strand sets and beta-sheet engineering utilizing chart calculations

Slide 8

Dataset and Statistics Extract proteins with high determination from Protein Data Bank (Berman et al., 2000) Use DSSP (Kabsch and Sander, 1983) to dole out intra-chain beta-sheet structure Use UniqueProt (Mika and Rost, 2003) to lessen excess Use PSI-BLAST (Altschul et al., 1997) to create profiles Statistics

Slide 9

Stage 1: Prediction of Beta-Residue Pairings Using 2D-RNN Target/Output Matrix (m×m) Input Matrix I (m×m) (i,j) 2D-RNN O = f(I) (i,j) T ij : 0/1 O ij : Pairing Prob. I ij X i - 2 X i - 1 X i X i +1 X i +2 X j - 2 X j - 1 X j X j +1 X j +2 |X i – X j | 20 profiles 3 SS 2 SA X i or X j is the position of beta-deposit i or j in the succession

Slide 10

An Example (Target) 1 2 3 4 5 6 7 Protein 1VJG Beta-Residue Pairing Map (Target Matrix)

Slide 11

An Example (Target) 1 2 3 4 5 6 7 Antiparallel Parallel Protein 1VJG Beta-Residue Pairing Map (Target Matrix)

Slide 12

An Example (Prediction)

Slide 13

Stage 2: Beta-Strand Alignment Antiparallel Use yield likelihood grid as scoring network Dynamic programming Disallow holes and utilize the streamlined inquiry calculation Parallel Total number of arrangements = 2(m+n-1)

Slide 14

Strand Alignment and Pairing Matrix The arrangement score is the total of the blending probabilities of the adjusted buildups The best arrangement is the arrangement with the most extreme score Strand Pairing Matrix Strand Pairing Matrix of 1VJG

Slide 15

Stage 3: Prediction of Beta-Strand Pairings and Beta-Sheet Architecture (Constraints) (a) Seven strands of protein 1VJG in grouping request (b) Beta-sheet topology of protein 1VJG

Slide 16

Stage 3: Prediction of Beta-Strand Pairings and Beta-Sheet Architecture (Constraints) (a) Seven strands of protein 1VJG in grouping request 3 accomplices Protein: 1B7G Rendered in Rasmol (b) Beta-sheet topology of protein 1VJG

Slide 17

Minimum Spanning Tree Like Algorithm Strand Pairing Graph (SPG) (a) Complete SPG Strand Pairing Matrix

Slide 18

Minimum Spanning Tree Like Algorithm Strand Pairing Graph (SPG) (b) True Weighted SPG (a) Complete SPG Strand Pairing Matrix Goal : Find an arrangement of associated subgraphs that augment the total of the arrangement scores and fulfill the limitations Algorithm : Minimum Spanning Tree Like Algorithm

Slide 19

An Example of MST Like Algorithm 1 2 3 4 5 6 7 Step 1: Pair strand 4 and 5 1 2 3 4 5 4 5 6 7 Strand Pairing Matrix of 1VJG

Slide 20

An Example of MST Like Algorithm 1 2 3 4 5 6 7 Step 2: Pair strand 1 and 2 1 2 3 4 5 4 5 6 7 2 1 Strand Pairing Matrix of 1VJG N

Slide 21

An Example of MST Like Algorithm 1 2 3 4 5 6 7 Step 3: Pair strand 1 and 3 1 2 3 4 5 4 5 6 7 2 1 3 Strand Pairing Matrix of 1VJG N

Slide 22

An Example of MST Like Algorithm 1 2 3 4 5 6 7 Step 4: Pair strand 3 and 6 1 2 3 4 5 4 5 6 7 2 1 3 6 Strand Pairing Matrix of 1VJG N

Slide 23

An Example of MST Like Algorithm 1 2 3 4 5 6 7 Step 5: Pair strand 6 and 7 1 2 3 4 5 4 5 6 C 7 2 1 3 6 7 Strand Pairing Matrix of 1VJG N

Slide 24

A New Fold Example (Last CASP) True auxiliary structure 1S12 (94 deposits) C EEEEE CCC EEEEE CCCCCHHHHHHHHHHHHHHHHHHHHCCC EEEEEE CC EEEEEE CCCCHHHHHHHHHHHHHHHHHHHHCCCC EEEEE CCCCCC Predicted optional structure by SSpro (Pollastri, et al., 2002) C EEEEEE CC EEEE CCCCCCCCHHHHHHHHHHHHHHHHHHHHHHH E HHCCCC EEEE HHHHHHHHHHHHHHHHHHHHHHHHHCCCC EEEEEEE CCC Strand Pairing Matrix Beta Sheet Topology 5 1 2 4 3 1s12 Rendered in Rasmol True: 1-2, 2-4, 3-4, 1-5 Predicted: 1-2, 2-4, 3-4, 4-5

Slide 25

Beta-Residue Pairing Results The precision of arbitrary calculation is 2.3%. ROC Plot

Slide 26

Strand Pairing Results Naïve calculation of blending every single contiguous strand Specificity = 42% Sensitivity = half All strand sets are nearby strand sets. MST like calculation Specificity = 53% Sensitivity = 59% >20% accurately anticipated strand sets are non-nearby strand sets.

Slide 27

Strand Alignment Results On the accurately anticipated strand matches On all local strand combines The precision of blending heading is 15% higher than that of the gauge calculation. The arrangement precision is essentially higher than past techniques.

Slide 28

Future Work and Applications Allow a cycle to handle beta-barrel, permit holes in arrangement for beta lump, include more sources of info (Punta and Rost, 2005) for beta buildup matching forecast Applications Contact outline acknowledgment Ab-initio structure expectation Model refinement Web server and dataset (SCRATCH suite) http://www.ics.uci.edu/~baldig/betasheet.html

Slide 29

Acknowledgment Pierre Baldi, Arlo Randall, Michael Sweredoski NIH give (LM-07443-01) NSF give (EIA-0321390)

SPONSORS