rosaura parisi ppt progetto
TRANSCRIPT
CNJEWDUHFYGRYBJDSKLHJUGFHGJBVUEDWHFYUGWIEGFIUWBJDSKLHJUGFHGJBVUEDWHFYUGWIEGFIUWBJDSKLHJUGFHGJCEDWHFYUGWIEGFIUWBJDSKLHJUGFHGJBVUCEDWHFYUGWIEGFIUWHEIGEOHMNVBICNJEWDUHFYGRYBJDSKLHJUGFHGJBVUEDWHFYUGWIEGFIUWBJDSKLHJUGFHGJBVUEDWHFYUGWIEGFIUWBJDSKLHJUGFHGJCEDWHFYUGWIEGFIUWBJDSKLHJUGFHGJBVUCEDWHFYUGWIEGFIUWHEIGEOHMNVBICNJEWDUHFYGRYBJDSKLHJUGFHGJBVUEDWHFYUGWIEGFIUWBJDSKLHJUGFHGJBVUEDWHFYUGWIEGFHGJBVUEDWHFYUGWIEGFIUWBJDSKLHJUGFHGJCEDWHFYUGWIEGFIUWBJDSKLHJUGFHGJBVUCEDWHFYUGWIEGFIUWHEIGEOHMNVBICNJEWDUHFYGRYBJDSKLHJUGFHGJBVUEDWHFYUGWIEGFIUWBJDSKLHJUGFHGJBVUEDWHFYUGWIEGFIUWBJDSKLHJUGFHGJCEDWHFYUGWIEGFIUWBJDSKLHJUGFHGJBVUCEDWHFYUGWIEGFIUWHEIGEOHMNVBICNJEWDUHFYGRYBJDSKLHJUGFHGJBVUEDWHFYUGWIEGFIUWBJDSKLHJUGFHGJBVUEDWHFYUGWIEGFHGJBVUEDWHFYUGWIEGFIUWBJDSKLHJUGFHGJCEDWHFYUGWIEGFIUWBJDSKLHJUGFHGJBVUCEDWHFYUGWIEGFIUWHEIGEOHMNVBICNJEWDUHFYGRYBJDSKLHJUGFHGJBVUEDWHFYUGWIEGFIUWBJDSKLHJUGFHGJBVUEDWHFYUGWIEGFIUWBJDSKLHJUGFHGJCEDWHFYUGWIEGFIUWBJDSKLHJUGFHGJBVUCEDWHFYUGWIEGFIUWHEIGEOHEGFIUEGFIUEGFIUEGFIUEGFIU
S e a rc h o f n o v e l m o l e c u l a rd e s c r i p t o rs fo r a nt i m i c ro b i a l p e p t i d e s
Dottorato di Ricerca in
Scienze del Farmaco
XVI ciclo
Tutor PhD candidateProf. Stefano Piotto Rosaura Parisi
Small peptides produced by multicellular organisms
12-50 amino acids in their L-configuration
α-helix structure
Amphipathic conformation
Positively charged
They are able to kill or to inhibit growth of various microorganisms
<3000 natural AMPs have been isolated and characterized from different sources
What are AMPs?
AMPs Limits Selectivity YADAMP QSAR GOALS PROTCOMP Validation Workflow
Advantages Disadvantages
Potential substitute of conventionalantibiotics
Protease susceptibility
They aren’t hindered by resistance High cost of production
Broad spectra of activity
Specificity
Limits of AMPs as therapeutic agents
AMPs Limits SelectivityYADAMP QSAR GOALS PROTCOMP Validation Workflow
Limits of current models
AMPs Limits YADAMP selectivityQSAR GOALS PROTCOMP Validation Workflow
Artificial Neural Networks
AMPs Limits SelectivityYADAMP QSAR GOALS PROTCOMP Validation Workflow
How to obtain reliable QSAR models?
An ANN with 10 hidden neurons was applied to 1000 antimicrobial peptides with less than 60 amino acids active against S. aureus
ANN model on AMPs active against S. aureus
Artificial Neural Networks (ANN)
AMPs Limits SelectivityYADAMP QSAR GOALS PROTCOMP Validation Workflow
Statistical evaluation of a GA model of AMP activity against S. aureus
Genetic Algorithms model
AMPs Limits SelectivityYADAMP QSAR GOALS PROTCOMP Validation Workflow
Training set: peptides with 6 < Lenght < 12 aa
MODEL2 performance
Condition
Total PopulationCondition
PositiveCondition Negative
Prevalence
86.52
Test outcome positive 543.9 22.1Precision
96.10
False discovery rate
3.90
Test outcome negative 137 84False omission rate
61.99
Negative predictive
value
38.01
Positive Likelihood Ratio
3.83
Sensitivity
79.88
False positive Rate
20.83
Accuracy
79.78
Negative Likelihood Ratio
0.25
False negative rate
20.12
Specificity
79.17
Diagnostic odd ratio
1508.99
Peptides are adapted to their lipid environment
AMPs Limits SelectivityYADAMP QSAR GOALS PROTCOMP Validation Workflow
eukaryoticmembrane
Prokaryoticmembrane
… better said: an example of a small portion of the lipid membrane of a particular organism
The selectivity depends on the “match”
My goal is to investigate any regularity between TMPs of
different organisms and see if they can be correlated with AMPs;
I will use the ‘distances’ among TMPs and AMPs as novel
descriptors in QSAR analysis
Main goals
AMPs Limits SelectivityYADAMP QSAR GOALS PROTCOMP Validation Workflow
How to improve QSAR models?
AAKKAAKKAAKK
IIIIIII-IIII
AAKKAAK-AAKKA
AMPs Limits SelectivityYADAMP QSAR GOALS PROTCOMP Validation Workflow
Define the variable that describes a biological entity (BE)
X = sequence (nucleotides, amino acids, ...) of the entity (such as a proteome)
Ci⊆X one possible substring of the sequence X (feature)
|Ci| occurrence number of characteristic
| X | = length of the sequence. Since | X | = Σ | x | ∀xCi
|Ci|/ | X | Weight feature
Define metrics
Calculate the distances among BEs
How it works
AMPs Limits SelectivityYADAMP QSAR goals PROTCOMP Validation Workflow
Validation 1: proteomes comparison
AMPs Limits SelectivityYADAMP QSAR goals PROTCOMP Validation Workflow
This work could allow the definition of new molecular descriptors for AMP based on the "distance"between the TMP of different organisms
Validation 2: TMPs comparison
AMPs Limits SelectivityYADAMP QSAR GOALS PROTCOMP Validation Workflow
Workflow
AMPs Limits SelectivityYADAMP QSAR goals PROTCOMP Validation Workflow
Work organization
Tasks I year II year III year
1 2 3 4 5 6
Selection of homogeneous AMP sets
QSAR analysis by means of GAs and ANNs
Definition of metrics within Protcomp
Definition and test of molecular descriptors
Design of AMPs with high selectivity
(Synthesis and test of peptides)
Stage abroad
AMPs Limits SelectivityYADAMP QSAR goals PROTCOMP Validation Workflow
> Thanks peptide in FASTA format
THANKSFORTHEATTENTION
Four models of interaction between cationic AMPs and cytoplasmic membrane
a) Barrel-stave pore
b) Carpet mechanism
c) Toroidal pore
d) Disordered toroidal pore
Mechanism of action: proposed models
Priddy, Kevin L., and Paul E. Keller. Artificial neural networks: an introduction. Vol. 68. SPIE Press, 2005.
PDB 2LGI PDB 4GU2
Red experimental 3D structure
Blue Predict 3D structure