Artificial Neural Networks Thomas Nordahl Petersen & Morten Nielsen
description
Transcript of Artificial Neural Networks Thomas Nordahl Petersen & Morten Nielsen
![Page 1: Artificial Neural Networks Thomas Nordahl Petersen & Morten Nielsen](https://reader036.fdocuments.in/reader036/viewer/2022062517/56813b57550346895da44989/html5/thumbnails/1.jpg)
Artificial Neural Networks
Thomas Nordahl Petersen &Morten Nielsen
![Page 2: Artificial Neural Networks Thomas Nordahl Petersen & Morten Nielsen](https://reader036.fdocuments.in/reader036/viewer/2022062517/56813b57550346895da44989/html5/thumbnails/2.jpg)
• A data-driven method to predict a feature, given a set of training data
• In biology input features could be amino acid sequence or nucleotides
• Secondary structure prediction
• Signal peptide prediction
• Surface accessibility
• Propeptide prediction
Use of artificial neural networks
N C
Signalpeptide
Propeptide Mature/active protein
![Page 3: Artificial Neural Networks Thomas Nordahl Petersen & Morten Nielsen](https://reader036.fdocuments.in/reader036/viewer/2022062517/56813b57550346895da44989/html5/thumbnails/3.jpg)
Neural network prediction methodshttp://www.cbs.dtu.dk/services/
![Page 4: Artificial Neural Networks Thomas Nordahl Petersen & Morten Nielsen](https://reader036.fdocuments.in/reader036/viewer/2022062517/56813b57550346895da44989/html5/thumbnails/4.jpg)
Pattern recognition
![Page 5: Artificial Neural Networks Thomas Nordahl Petersen & Morten Nielsen](https://reader036.fdocuments.in/reader036/viewer/2022062517/56813b57550346895da44989/html5/thumbnails/5.jpg)
Biological Neural network
![Page 6: Artificial Neural Networks Thomas Nordahl Petersen & Morten Nielsen](https://reader036.fdocuments.in/reader036/viewer/2022062517/56813b57550346895da44989/html5/thumbnails/6.jpg)
Biological neuron structure
Synapse
Neuron
TerminalSeveral connections
![Page 7: Artificial Neural Networks Thomas Nordahl Petersen & Morten Nielsen](https://reader036.fdocuments.in/reader036/viewer/2022062517/56813b57550346895da44989/html5/thumbnails/7.jpg)
Diversity of interactions in a network enables complex calculations
• Similar in biological and artificial systems
• Excitatory (+) and inhibitory (-) relations between compute units
fire0
1
![Page 8: Artificial Neural Networks Thomas Nordahl Petersen & Morten Nielsen](https://reader036.fdocuments.in/reader036/viewer/2022062517/56813b57550346895da44989/html5/thumbnails/8.jpg)
Transfer of biological principles to artificial neural network algorithms
• Non-linear relation between input and output
• Massively parallel information processing
• Data-driven construction of algorithms
• Ability to generalize to new data items
![Page 9: Artificial Neural Networks Thomas Nordahl Petersen & Morten Nielsen](https://reader036.fdocuments.in/reader036/viewer/2022062517/56813b57550346895da44989/html5/thumbnails/9.jpg)
Sparse encoding of amino acid sequence windows
![Page 10: Artificial Neural Networks Thomas Nordahl Petersen & Morten Nielsen](https://reader036.fdocuments.in/reader036/viewer/2022062517/56813b57550346895da44989/html5/thumbnails/10.jpg)
Sparse encoding
Inp Neuron 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
AAcid
A 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
R 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
N 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
D 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
C 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
Q 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0
E 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0
![Page 11: Artificial Neural Networks Thomas Nordahl Petersen & Morten Nielsen](https://reader036.fdocuments.in/reader036/viewer/2022062517/56813b57550346895da44989/html5/thumbnails/11.jpg)
BLOSUM encoding (Blosum50 matrix)
A R N D C Q E G H I L K M F P S T W Y V A 4 -1 -2 -2 0 -1 -1 0 -2 -1 -1 -1 -1 -2 -1 1 0 -3 -2 0R -1 5 0 -2 -3 1 0 -2 0 -3 -2 2 -1 -3 -2 -1 -1 -3 -2 -3N -2 0 6 1 -3 0 0 0 1 -3 -3 0 -2 -3 -2 1 0 -4 -2 -3D -2 -2 1 6 -3 0 2 -1 -1 -3 -4 -1 -3 -3 -1 0 -1 -4 -3 -3C 0 -3 -3 -3 9 -3 -4 -3 -3 -1 -1 -3 -1 -2 -3 -1 -1 -2 -2 -1Q -1 1 0 0 -3 5 2 -2 0 -3 -2 1 0 -3 -1 0 -1 -2 -1 -2E -1 0 0 2 -4 2 5 -2 0 -3 -3 1 -2 -3 -1 0 -1 -3 -2 -2G 0 -2 0 -1 -3 -2 -2 6 -2 -4 -4 -2 -3 -3 -2 0 -2 -2 -3 -3H -2 0 1 -1 -3 0 0 -2 8 -3 -3 -1 -2 -1 -2 -1 -2 -2 2 -3I -1 -3 -3 -3 -1 -3 -3 -4 -3 4 2 -3 1 0 -3 -2 -1 -3 -1 3L -1 -2 -3 -4 -1 -2 -3 -4 -3 2 4 -2 2 0 -3 -2 -1 -2 -1 1K -1 2 0 -1 -3 1 1 -2 -1 -3 -2 5 -1 -3 -1 0 -1 -3 -2 -2M -1 -1 -2 -3 -1 0 -2 -3 -2 1 2 -1 5 0 -2 -1 -1 -1 -1 1F -2 -3 -3 -3 -2 -3 -3 -3 -1 0 0 -3 0 6 -4 -2 -2 1 3 -1P -1 -2 -2 -1 -3 -1 -1 -2 -2 -3 -3 -1 -2 -4 7 -1 -1 -4 -3 -2S 1 -1 1 0 -1 0 0 0 -1 -2 -2 0 -1 -2 -1 4 1 -3 -2 -2T 0 -1 0 -1 -1 -1 -1 -2 -2 -1 -1 -1 -1 -2 -1 1 5 -2 -2 0W -3 -3 -4 -4 -2 -2 -3 -2 -2 -3 -2 -3 -1 1 -4 -3 -2 11 2 -3Y -2 -2 -2 -3 -2 -1 -2 -3 2 -1 -1 -2 -1 3 -3 -2 -2 2 7 -1V 0 -3 -3 -3 -1 -2 -2 -3 -3 3 1 -2 1 -1 -2 -2 0 -3 -1 4
![Page 12: Artificial Neural Networks Thomas Nordahl Petersen & Morten Nielsen](https://reader036.fdocuments.in/reader036/viewer/2022062517/56813b57550346895da44989/html5/thumbnails/12.jpg)
Sequence encoding (continued)
• Sparse encoding
– V:0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1
– L:0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0
– V.L=0 (unrelated)
• Blosum encoding
– V: 0 -3 -3 -3 -1 -2 -2 -3 -3 3 1 -2 1 -1 -2 -2 0 -3 -1 4
– L:-1 -2 -3 -4 -1 -2 -3 -4 -3 2 4 -2 2 0 -3 -2 -1 -2 -1 1
– V.L = 0.88 (highly related)
– V.R = -0.08 (close to unrelated)
![Page 13: Artificial Neural Networks Thomas Nordahl Petersen & Morten Nielsen](https://reader036.fdocuments.in/reader036/viewer/2022062517/56813b57550346895da44989/html5/thumbnails/13.jpg)
I1 I2 I3
h1 h2
O1
Input
h1 h1
hidden
output
= 1/ (1+e-x)
o=H1*v1,1 + H2*v2,1O1 = (o)
w1,1
v1,1
w3,1
v2,1
Error = O - True
w1,2
![Page 14: Artificial Neural Networks Thomas Nordahl Petersen & Morten Nielsen](https://reader036.fdocuments.in/reader036/viewer/2022062517/56813b57550346895da44989/html5/thumbnails/14.jpg)
Sigmodial or logistic function
![Page 15: Artificial Neural Networks Thomas Nordahl Petersen & Morten Nielsen](https://reader036.fdocuments.in/reader036/viewer/2022062517/56813b57550346895da44989/html5/thumbnails/15.jpg)
![Page 16: Artificial Neural Networks Thomas Nordahl Petersen & Morten Nielsen](https://reader036.fdocuments.in/reader036/viewer/2022062517/56813b57550346895da44989/html5/thumbnails/16.jpg)
Training and error reduction
![Page 17: Artificial Neural Networks Thomas Nordahl Petersen & Morten Nielsen](https://reader036.fdocuments.in/reader036/viewer/2022062517/56813b57550346895da44989/html5/thumbnails/17.jpg)
Training and error reduction
![Page 18: Artificial Neural Networks Thomas Nordahl Petersen & Morten Nielsen](https://reader036.fdocuments.in/reader036/viewer/2022062517/56813b57550346895da44989/html5/thumbnails/18.jpg)
Training and error reduction
Size matters
![Page 19: Artificial Neural Networks Thomas Nordahl Petersen & Morten Nielsen](https://reader036.fdocuments.in/reader036/viewer/2022062517/56813b57550346895da44989/html5/thumbnails/19.jpg)
ß-strand
Helix
TurnBend
Secondary Structure Elements
![Page 20: Artificial Neural Networks Thomas Nordahl Petersen & Morten Nielsen](https://reader036.fdocuments.in/reader036/viewer/2022062517/56813b57550346895da44989/html5/thumbnails/20.jpg)
Neural Network Architecture
IKEEHVI IQAE
HEC
IKEEHVIIQAEFYLNPDQSGEF…..Window
Input Layer
Hidden Layer
Output Layer
Weights
![Page 21: Artificial Neural Networks Thomas Nordahl Petersen & Morten Nielsen](https://reader036.fdocuments.in/reader036/viewer/2022062517/56813b57550346895da44989/html5/thumbnails/21.jpg)
• Normally the best prediction is obtained by averaging• results from several predictions - “wisdom of the crowd
• Two types of neural networks• Prediction of features in classes/bins e.g. H, E or C (1,0,0)
• Values close to 1 or 0 are more accurate than values close to 1/2
• Prediction of real values e.g. Surface accessibility (0.43)• Reliability of a prediction is more difficult to estimate
Predictions and reliability of a prediction
![Page 22: Artificial Neural Networks Thomas Nordahl Petersen & Morten Nielsen](https://reader036.fdocuments.in/reader036/viewer/2022062517/56813b57550346895da44989/html5/thumbnails/22.jpg)
![Page 23: Artificial Neural Networks Thomas Nordahl Petersen & Morten Nielsen](https://reader036.fdocuments.in/reader036/viewer/2022062517/56813b57550346895da44989/html5/thumbnails/23.jpg)
Eukaryotic SP & TM
Signal peptide cleavage1523 seq
C-terminal end ofTM-regions669 seq
![Page 24: Artificial Neural Networks Thomas Nordahl Petersen & Morten Nielsen](https://reader036.fdocuments.in/reader036/viewer/2022062517/56813b57550346895da44989/html5/thumbnails/24.jpg)
Signal peptide prediction
Signal pepdide likenessCleavage siteCombined information
![Page 25: Artificial Neural Networks Thomas Nordahl Petersen & Morten Nielsen](https://reader036.fdocuments.in/reader036/viewer/2022062517/56813b57550346895da44989/html5/thumbnails/25.jpg)
Propeptide predictionMany secretory proteins and peptides are synthesized as inactive precursors that inaddition to signal peptide cleavage undergo post-translational processing to becomebiologically active polypeptides. Precursors are usually cleaved at sites composed ofsingle or paired basic amino acid residues by members of the subtilisin/kexin-likeproprotein convertase (PC) family. In mammals, seven members have been identified,with furin being the one first discovered and best characterized. Recently, theinvolvement of furin in diseases ranging from Alzheimer's disease and cancer toanthrax and Ebola fever has created additional focus on proprotein processing.We have developed a method for prediction of cleavage sites for PCs based onartificial neural networks. Two different types of neural networks have beenconstructed: a furin-specific network based on experimental results derived fromthe literature, and a general PC-specific network trained on data from the Swiss-Protprotein database. The method predicts cleavage sites in independent sequences witha sensitivity of 95% for the furin neural network and 62% for the general PC network.
Protein Engineering, Design and Selection: 17: 107-112, 2004.
General cleavage: R/K-Xn-R/K , n=0, 2, 4, 6
Furin cleavage: R-X-R/K-R
![Page 26: Artificial Neural Networks Thomas Nordahl Petersen & Morten Nielsen](https://reader036.fdocuments.in/reader036/viewer/2022062517/56813b57550346895da44989/html5/thumbnails/26.jpg)
Propeptide prediction
Furin cleavage
![Page 27: Artificial Neural Networks Thomas Nordahl Petersen & Morten Nielsen](https://reader036.fdocuments.in/reader036/viewer/2022062517/56813b57550346895da44989/html5/thumbnails/27.jpg)
![Page 28: Artificial Neural Networks Thomas Nordahl Petersen & Morten Nielsen](https://reader036.fdocuments.in/reader036/viewer/2022062517/56813b57550346895da44989/html5/thumbnails/28.jpg)
![Page 29: Artificial Neural Networks Thomas Nordahl Petersen & Morten Nielsen](https://reader036.fdocuments.in/reader036/viewer/2022062517/56813b57550346895da44989/html5/thumbnails/29.jpg)