1) Computer-aided drug design uses computational techniques to aid in the drug discovery process, including finding and storing relevant information, modeling existing lead compounds, and developing new lead compounds.
2) Key techniques include pharmacophore modeling to identify functional groups important for activity, 3D QSAR to develop quantitative structure-activity models, and docking to model interactions of ligands with protein targets.
3) Developing new leads can involve de novo design to build ligands into a target structure, database searching using pharmacophore queries, and combinatorial library design to rapidly screen many potential compounds.
1. The document describes a study using spectroscopy to detect cervical dysplasia. Non-negative matrix factorization (NNMF) was used to decompose spectroscopy data into constituent source spectra and concentrations.
2. A machine learning model (Lasso regression) combined NNMF source concentrations to predict dysplasia levels. This improved prediction performance over individual reflectance or fluorescence data.
3. Two-dimensional disease maps were created locating cervical dysplasia tissue using the machine learning results. These maps correctly identified biopsy-confirmed normal and dysplastic tissue locations.
IRJET- Semantic Retrieval of Trademarks based on Text and Images Conceptu...IRJET Journal
The document proposes a novel Weakly-supervised Deep Matrix Factorization (WDMF) algorithm for social image tag refinement, assignment and retrieval. WDMF uncovers latent image and tag representations in a latent subspace by exploiting weakly supervised tagging information, visual structure and semantic structure. It can handle noisy, incomplete or subjective tags and noisy or redundant visual features. An optimization problem with a well-defined objective function is formulated and solved using gradient descent with curvilinear search. Extensive experiments on two real-world social image databases demonstrate the effectiveness of the approach.
The GEM (General Enterprise Management) methodology provides an ontology and categories for modeling an enterprise. It defines object categories like locations, organizations, functions, processes, and resources. It also defines relation types like containment, sequence, reference, and change. Primary assertions use these categories and relations to model facts about how objects in the enterprise are related, such as an organization being located somewhere, performing functions, and using resources. The methodology also outlines a management life cycle for applying the modeling to functions like intelligence management, enterprise operations management, and mission management.
1) Computer-aided drug design uses computational techniques to aid in the drug discovery process, including finding and storing relevant information, modeling existing lead compounds, and developing new lead compounds.
2) Key techniques include pharmacophore modeling to identify functional groups important for activity, 3D QSAR to develop quantitative structure-activity models, and docking to model interactions of ligands with protein targets.
3) Developing new leads can involve de novo design to build ligands into a target structure, database searching using pharmacophore queries, and combinatorial library design to rapidly screen many potential compounds.
A soldier named Chicaiza Cristian listed his free time activities as meeting friends, going to the beach to relax, playing volleyball, and sleeping. The following morning he would return home after completing his training at the Army Soldiers Training School.
1. Computer-aided drug design uses computational techniques to aid in the drug discovery process, including finding and storing relevant information, modeling existing lead compounds, and developing new lead compounds.
2. Key techniques include pharmacophore modeling to identify functional groups important for activity, quantitative structure-activity relationship modeling to predict activity, molecular docking to study binding, and free energy perturbation calculations to compare binding of candidates.
3. The workflow involves generating working models of ligands or targets, proposing new lead structures through techniques like de novo design or database searching, and evaluating candidates through synthesis and testing.
1. The document describes a study using spectroscopy to detect cervical dysplasia. Non-negative matrix factorization (NNMF) was used to decompose spectroscopy data into constituent source spectra and concentrations.
2. A machine learning model (Lasso regression) combined NNMF source concentrations to predict dysplasia levels. This improved prediction performance over individual reflectance or fluorescence data.
3. Two-dimensional disease maps were created locating cervical dysplasia tissue using the machine learning results. These maps correctly identified biopsy-confirmed normal and dysplastic tissue locations.
IRJET- Semantic Retrieval of Trademarks based on Text and Images Conceptu...IRJET Journal
The document proposes a novel Weakly-supervised Deep Matrix Factorization (WDMF) algorithm for social image tag refinement, assignment and retrieval. WDMF uncovers latent image and tag representations in a latent subspace by exploiting weakly supervised tagging information, visual structure and semantic structure. It can handle noisy, incomplete or subjective tags and noisy or redundant visual features. An optimization problem with a well-defined objective function is formulated and solved using gradient descent with curvilinear search. Extensive experiments on two real-world social image databases demonstrate the effectiveness of the approach.
The GEM (General Enterprise Management) methodology provides an ontology and categories for modeling an enterprise. It defines object categories like locations, organizations, functions, processes, and resources. It also defines relation types like containment, sequence, reference, and change. Primary assertions use these categories and relations to model facts about how objects in the enterprise are related, such as an organization being located somewhere, performing functions, and using resources. The methodology also outlines a management life cycle for applying the modeling to functions like intelligence management, enterprise operations management, and mission management.
1) Computer-aided drug design uses computational techniques to aid in the drug discovery process, including finding and storing relevant information, modeling existing lead compounds, and developing new lead compounds.
2) Key techniques include pharmacophore modeling to identify functional groups important for activity, 3D QSAR to develop quantitative structure-activity models, and docking to model interactions of ligands with protein targets.
3) Developing new leads can involve de novo design to build ligands into a target structure, database searching using pharmacophore queries, and combinatorial library design to rapidly screen many potential compounds.
A soldier named Chicaiza Cristian listed his free time activities as meeting friends, going to the beach to relax, playing volleyball, and sleeping. The following morning he would return home after completing his training at the Army Soldiers Training School.
1. Computer-aided drug design uses computational techniques to aid in the drug discovery process, including finding and storing relevant information, modeling existing lead compounds, and developing new lead compounds.
2. Key techniques include pharmacophore modeling to identify functional groups important for activity, quantitative structure-activity relationship modeling to predict activity, molecular docking to study binding, and free energy perturbation calculations to compare binding of candidates.
3. The workflow involves generating working models of ligands or targets, proposing new lead structures through techniques like de novo design or database searching, and evaluating candidates through synthesis and testing.
This document discusses the use of monoclonal antibodies for cancer therapy. It provides background on conventional chemotherapy and highlights limitations. It then covers the history and development of monoclonal antibodies, including their production and mechanisms of targeting cancer cells, such as antigen cross-linking, activating death receptors, and delivering cytotoxic agents. Specific examples of toxin-immunoconjugates and antibody-directed enzyme prodrug therapy are described. The monoclonal antibody Rituximab is discussed as the first FDA-approved therapeutic monoclonal antibody for cancer.
This document discusses the use of monoclonal antibodies for cancer therapy. It provides background on conventional chemotherapy and highlights limitations. It then covers the history and development of monoclonal antibodies, including their production and mechanisms of targeting cancer cells through antigen cross-linking, activating death receptors, or delivering cytotoxic agents. Specific examples of toxin-immunoconjugates and antibody-directed enzyme prodrug therapy are described. The monoclonal antibody Rituximab is discussed as the first FDA-approved therapeutic monoclonal antibody for cancer.
1) Computer-aided drug design uses computational techniques to aid in the drug discovery process, including finding and storing relevant information, modeling existing lead compounds, and developing new lead compounds.
2) Key techniques include pharmacophore modeling to identify functional groups important for activity, 3D QSAR to develop quantitative structure-activity models, and docking to model interactions of ligands with protein targets.
3) Developing new leads can involve de novo design to build ligands into a target structure, database searching using pharmacophore queries, and combinatorial library design to rapidly screen many potential compounds.
Toxicogenomics uses genomic technologies to study the effects of toxicants like drugs and chemicals on human health. It provides information on their molecular-level effects and potential toxicities. While this field shows promise to enhance risk assessments, more coordinated efforts are needed to generate data, study existing data in new ways, and address challenges. A large public database and initiatives like a proposed Human Toxicogenomics Initiative could help realize its potential to improve predictive toxicology and public health decisions.
The document promotes Ditch Witch vacuum excavation systems for utilities and water departments. It highlights the technological and safety advancements of Ditch Witch products that can benefit agencies. Examples are provided of how Ditch Witch vacuums can efficiently excavate work spaces and complete jobs faster than traditional methods. Key advantages over competitors like its quieter operation and fully enclosed components are also outlined.
This document discusses the use of monoclonal antibodies for cancer therapy. It provides background on conventional chemotherapy and highlights limitations. It then covers the history and development of monoclonal antibodies, including their production and mechanisms of targeting cancer cells, such as antigen cross-linking, activating death receptors, and delivering cytotoxic agents. Specific examples of toxin-immunoconjugates and antibody-directed enzyme prodrug therapy are described. The mechanism and applications of the monoclonal antibody Rituximab for lymphoma are discussed. In conclusion, the document notes the potential for optimizing monoclonal antibody combinations with chemotherapy and radiation therapy.
This document discusses compensation and employee remuneration. It defines compensation as all forms of financial returns, benefits, and services employees receive as part of their employment. Compensation includes wages, salaries, bonuses, benefits like health insurance, and paid time off. High compensation and appealing benefits help attract and retain employees. The compensation package at Ashok Leyland includes direct financial compensation like salary and bonuses, as well as indirect financial and non-financial benefits. The goals of compensation are to attract, motivate, and retain capable employees. Compensation impacts performance and is based on theories like equity and expectancy theories.
Toxicogenomics uses gene expression profiling to study how organisms respond to toxic compounds on a global scale. This new approach promises to greatly advance toxicology research. It may help identify toxic mechanisms earlier and assist in predicting compound toxicity. Challenges include interpreting large gene expression datasets and linking changes to specific toxic effects. Progress has been made using toxicogenomics to predict compound mode of action and toxicity pathways. Integrating gene expression data with traditional toxicology can help realize the full potential of this new approach.
II-PIC 2017: Drug Discovery of Novel Molecules using Chemical Data Mining toolDr. Haxel Consult
Muthukumarasamy Karthikeyan (CSIR-National Chemical Laboratory, India)
Surojit Sadhu (Advent Informatics, India)
Virtual screening (VS) and chemical data extracted from evidence based sources are the backbone of computational drug discovery workflow, an indispensable component in all drug design programs. It involves a host of modelling techniques from simple similarity search methods to advanced algorithms for finding the accurate bioactive conformation of a molecule to bind to its corresponding target. Chemoinformatics supports virtual screening at multiple levels during the lead optimization stage by suggesting suitable filters for numerous screenings by utilizing the power of data integration from multiple sources and derived knowledge that is essential for decision support in drug discovery and development. It is therefore pertinent to develop tools, data and emerging methods in chemoinformatics to fully understand their role and applications in virtual screening. Recently we have developed chemical informatics tools to assist drug discovery by chemical data extraction from literature, virtual library design, analysis and screening methods on selected case studies.
Higher order thinking skills are referred to in many of the new education and economic papers but what are they ? This diagram explains it and really shows that if we are taking an enterprising approach to learning - making it real, relevant and pupil centred higher order thinking skills will be developed through the process. .
This document outlines Bloom's Revised Taxonomy, which categorizes levels of thinking skills from lower-order to higher-order. It lists actions and products for each of the six major cognitive process categories - Remembering, Understanding, Applying, Analyzing, Evaluating, and Creating. These categories move from simpler recall tasks to more complex skills such as synthesis, evaluation and creation. Learning activities are provided as examples for each level.
Rahul Biswas is a computational physicist working as a postdoctoral research associate at the Center for Gravitational Wave and Astronomy. He obtained his PhD in Physics from the University of Wisconsin Milwaukee in 2010. His current research involves analyzing 250GB of astrophysical data from LIGO and Virgo detectors to classify noise transients and understand their origins using techniques like time series analysis, machine learning algorithms, and statistical modeling. Previously as a research assistant, he performed data analysis of LIGO-Virgo experiments to search for gravitational wave sources.
Tracking Trends in Korean Information Science Research, 2000-2011SoYoung YU
This is a presentation file of "Tracking Trends in Korean Information Science Research, 2000-2011" which was published in COLLNET 2012 proceeding, October 23rd, 2012.
If you need a full paper of it, feel free to contact So Young Yu (soyoung.yu21@gmail.com)
Byte is a unit of digital information that typically consists of eight bits. Historically, a byte represented the number of bits used to encode a single character. It is the basic addressable element in many computer architectures.
Mind the Byte is a consultancy that provides computational scientific solutions for researchers. They can handle all scientific computation needs, from large data analysis to drug discovery and prediction of compound activity. Their services include drug discovery, molecular modelling, support, education, and cloud computing.
Integrating Public and Private Data: Lessons Learned from UnisonReece Hart
The document discusses lessons learned from integrating public and private data using the Unison platform. It describes the types of data that can be integrated, including genomics, proteomics, chemistry, networks, and clinical data. It outlines different types of integration like semantic and source integration. Challenges of integration include establishing relationships between data and handling frequent updates. Benefits include enabling analysis across diverse data types and centralizing data. Unison integrates sequences, annotations, auxiliary data and precomputed predictions from sources like UniProt and Ensembl to power applications, in-house tools and data mining projects.
Data Integration at the Ontology Engineering GroupOscar Corcho
Presentation done on the work being done on Data Integration at OEG-UPM (http://www.oeg-upm.net/), for the CredIBLE workshop, in Sophia-Antipolis (October 15th, 2012).
This lab aims to analyze gene expression data from a study on the response of human fibroblasts to serum. The study used cDNA microarrays to explore the temporal program of gene expression during this physiological response, identifying genes clustered by their expression patterns. Many features of the transcriptional program appeared related to wound repair processes, suggesting fibroblasts play a richer role than previously thought. The lab will introduce gene expression analysis, demonstrate basic Excel tools for working with microarray data, and use the GEPAS suite to apply the full microarray analysis process to the fibroblast dataset, including preprocessing, clustering, and identifying differentially expressed genes.
The document summarizes a research paper that proposed a link prediction model for citation networks. It applied support vector machines (SVMs) as the classifier and used 11 features optimized for citation networks across 5 academic fields. The model was able to better predict links compared to just using the classifier's performance metrics. However, the effective features varied by academic field, suggesting different models should be applied for different research areas.
The document discusses the ISA (Investigation/Study/Assay) framework for enabling data reuse and reproducibility in bioscience research. The ISA framework provides a generic format for rich experimental descriptions and an infrastructure of open source software tools. It aims to minimize the burden of reporting, curating, sharing data and metadata from bioscience experiments to enable comprehension, reuse of data, and reproducibility. The framework promotes community engagement to develop community standards and document use cases.
Berlin center for genome based bioinformatics koch05Slava Karpov
This document summarizes the research activities of the Berlin Center for Genome Based Bioinformatics at the Technical University of Applied Sciences. The center focuses on modeling and analyzing biochemical systems using Petri nets. Specifically, it has modeled central metabolic pathways like glycolysis and developed Petri net tools to validate biochemical networks and analyze the behavior of large networks like E. coli metabolism.
This document discusses the use of monoclonal antibodies for cancer therapy. It provides background on conventional chemotherapy and highlights limitations. It then covers the history and development of monoclonal antibodies, including their production and mechanisms of targeting cancer cells, such as antigen cross-linking, activating death receptors, and delivering cytotoxic agents. Specific examples of toxin-immunoconjugates and antibody-directed enzyme prodrug therapy are described. The monoclonal antibody Rituximab is discussed as the first FDA-approved therapeutic monoclonal antibody for cancer.
This document discusses the use of monoclonal antibodies for cancer therapy. It provides background on conventional chemotherapy and highlights limitations. It then covers the history and development of monoclonal antibodies, including their production and mechanisms of targeting cancer cells through antigen cross-linking, activating death receptors, or delivering cytotoxic agents. Specific examples of toxin-immunoconjugates and antibody-directed enzyme prodrug therapy are described. The monoclonal antibody Rituximab is discussed as the first FDA-approved therapeutic monoclonal antibody for cancer.
1) Computer-aided drug design uses computational techniques to aid in the drug discovery process, including finding and storing relevant information, modeling existing lead compounds, and developing new lead compounds.
2) Key techniques include pharmacophore modeling to identify functional groups important for activity, 3D QSAR to develop quantitative structure-activity models, and docking to model interactions of ligands with protein targets.
3) Developing new leads can involve de novo design to build ligands into a target structure, database searching using pharmacophore queries, and combinatorial library design to rapidly screen many potential compounds.
Toxicogenomics uses genomic technologies to study the effects of toxicants like drugs and chemicals on human health. It provides information on their molecular-level effects and potential toxicities. While this field shows promise to enhance risk assessments, more coordinated efforts are needed to generate data, study existing data in new ways, and address challenges. A large public database and initiatives like a proposed Human Toxicogenomics Initiative could help realize its potential to improve predictive toxicology and public health decisions.
The document promotes Ditch Witch vacuum excavation systems for utilities and water departments. It highlights the technological and safety advancements of Ditch Witch products that can benefit agencies. Examples are provided of how Ditch Witch vacuums can efficiently excavate work spaces and complete jobs faster than traditional methods. Key advantages over competitors like its quieter operation and fully enclosed components are also outlined.
This document discusses the use of monoclonal antibodies for cancer therapy. It provides background on conventional chemotherapy and highlights limitations. It then covers the history and development of monoclonal antibodies, including their production and mechanisms of targeting cancer cells, such as antigen cross-linking, activating death receptors, and delivering cytotoxic agents. Specific examples of toxin-immunoconjugates and antibody-directed enzyme prodrug therapy are described. The mechanism and applications of the monoclonal antibody Rituximab for lymphoma are discussed. In conclusion, the document notes the potential for optimizing monoclonal antibody combinations with chemotherapy and radiation therapy.
This document discusses compensation and employee remuneration. It defines compensation as all forms of financial returns, benefits, and services employees receive as part of their employment. Compensation includes wages, salaries, bonuses, benefits like health insurance, and paid time off. High compensation and appealing benefits help attract and retain employees. The compensation package at Ashok Leyland includes direct financial compensation like salary and bonuses, as well as indirect financial and non-financial benefits. The goals of compensation are to attract, motivate, and retain capable employees. Compensation impacts performance and is based on theories like equity and expectancy theories.
Toxicogenomics uses gene expression profiling to study how organisms respond to toxic compounds on a global scale. This new approach promises to greatly advance toxicology research. It may help identify toxic mechanisms earlier and assist in predicting compound toxicity. Challenges include interpreting large gene expression datasets and linking changes to specific toxic effects. Progress has been made using toxicogenomics to predict compound mode of action and toxicity pathways. Integrating gene expression data with traditional toxicology can help realize the full potential of this new approach.
II-PIC 2017: Drug Discovery of Novel Molecules using Chemical Data Mining toolDr. Haxel Consult
Muthukumarasamy Karthikeyan (CSIR-National Chemical Laboratory, India)
Surojit Sadhu (Advent Informatics, India)
Virtual screening (VS) and chemical data extracted from evidence based sources are the backbone of computational drug discovery workflow, an indispensable component in all drug design programs. It involves a host of modelling techniques from simple similarity search methods to advanced algorithms for finding the accurate bioactive conformation of a molecule to bind to its corresponding target. Chemoinformatics supports virtual screening at multiple levels during the lead optimization stage by suggesting suitable filters for numerous screenings by utilizing the power of data integration from multiple sources and derived knowledge that is essential for decision support in drug discovery and development. It is therefore pertinent to develop tools, data and emerging methods in chemoinformatics to fully understand their role and applications in virtual screening. Recently we have developed chemical informatics tools to assist drug discovery by chemical data extraction from literature, virtual library design, analysis and screening methods on selected case studies.
Higher order thinking skills are referred to in many of the new education and economic papers but what are they ? This diagram explains it and really shows that if we are taking an enterprising approach to learning - making it real, relevant and pupil centred higher order thinking skills will be developed through the process. .
This document outlines Bloom's Revised Taxonomy, which categorizes levels of thinking skills from lower-order to higher-order. It lists actions and products for each of the six major cognitive process categories - Remembering, Understanding, Applying, Analyzing, Evaluating, and Creating. These categories move from simpler recall tasks to more complex skills such as synthesis, evaluation and creation. Learning activities are provided as examples for each level.
Rahul Biswas is a computational physicist working as a postdoctoral research associate at the Center for Gravitational Wave and Astronomy. He obtained his PhD in Physics from the University of Wisconsin Milwaukee in 2010. His current research involves analyzing 250GB of astrophysical data from LIGO and Virgo detectors to classify noise transients and understand their origins using techniques like time series analysis, machine learning algorithms, and statistical modeling. Previously as a research assistant, he performed data analysis of LIGO-Virgo experiments to search for gravitational wave sources.
Tracking Trends in Korean Information Science Research, 2000-2011SoYoung YU
This is a presentation file of "Tracking Trends in Korean Information Science Research, 2000-2011" which was published in COLLNET 2012 proceeding, October 23rd, 2012.
If you need a full paper of it, feel free to contact So Young Yu (soyoung.yu21@gmail.com)
Byte is a unit of digital information that typically consists of eight bits. Historically, a byte represented the number of bits used to encode a single character. It is the basic addressable element in many computer architectures.
Mind the Byte is a consultancy that provides computational scientific solutions for researchers. They can handle all scientific computation needs, from large data analysis to drug discovery and prediction of compound activity. Their services include drug discovery, molecular modelling, support, education, and cloud computing.
Integrating Public and Private Data: Lessons Learned from UnisonReece Hart
The document discusses lessons learned from integrating public and private data using the Unison platform. It describes the types of data that can be integrated, including genomics, proteomics, chemistry, networks, and clinical data. It outlines different types of integration like semantic and source integration. Challenges of integration include establishing relationships between data and handling frequent updates. Benefits include enabling analysis across diverse data types and centralizing data. Unison integrates sequences, annotations, auxiliary data and precomputed predictions from sources like UniProt and Ensembl to power applications, in-house tools and data mining projects.
Data Integration at the Ontology Engineering GroupOscar Corcho
Presentation done on the work being done on Data Integration at OEG-UPM (http://www.oeg-upm.net/), for the CredIBLE workshop, in Sophia-Antipolis (October 15th, 2012).
This lab aims to analyze gene expression data from a study on the response of human fibroblasts to serum. The study used cDNA microarrays to explore the temporal program of gene expression during this physiological response, identifying genes clustered by their expression patterns. Many features of the transcriptional program appeared related to wound repair processes, suggesting fibroblasts play a richer role than previously thought. The lab will introduce gene expression analysis, demonstrate basic Excel tools for working with microarray data, and use the GEPAS suite to apply the full microarray analysis process to the fibroblast dataset, including preprocessing, clustering, and identifying differentially expressed genes.
The document summarizes a research paper that proposed a link prediction model for citation networks. It applied support vector machines (SVMs) as the classifier and used 11 features optimized for citation networks across 5 academic fields. The model was able to better predict links compared to just using the classifier's performance metrics. However, the effective features varied by academic field, suggesting different models should be applied for different research areas.
The document discusses the ISA (Investigation/Study/Assay) framework for enabling data reuse and reproducibility in bioscience research. The ISA framework provides a generic format for rich experimental descriptions and an infrastructure of open source software tools. It aims to minimize the burden of reporting, curating, sharing data and metadata from bioscience experiments to enable comprehension, reuse of data, and reproducibility. The framework promotes community engagement to develop community standards and document use cases.
Berlin center for genome based bioinformatics koch05Slava Karpov
This document summarizes the research activities of the Berlin Center for Genome Based Bioinformatics at the Technical University of Applied Sciences. The center focuses on modeling and analyzing biochemical systems using Petri nets. Specifically, it has modeled central metabolic pathways like glycolysis and developed Petri net tools to validate biochemical networks and analyze the behavior of large networks like E. coli metabolism.
Cadd and molecular modeling for M.PharmShikha Popali
THE CADD IS FOR THE DRUG DEVELOPMENT THE DIFFERENT STRATEGIES ARE MENTIONED LIKE QSAR MOLECULAR DOCKING, THE DIFFERENT DIMNSIONAL FORMS OF QSAR , THE ADVANCE SAR of it.
This document describes a new immune-inspired algorithm called IMSA for multiple sequence alignment of proteins. IMSA incorporates strategies to create an initial population and specific mutation operators. It uses the weighted sum of pairs as an objective function to evaluate candidate alignments. The algorithm was tested on benchmarks from BALIBASE and was found to produce alignments comparable to state-of-the-art methods, while also generating multiple suboptimal alignments. This allows assessment of biologically relevant alignments.
The document presents a taxonomy-based approach for designing glyphs to visualize workflows of biological experiments. It develops a taxonomy to systematically organize concepts related to biological experiments. It orders visual channels like color, shape and size based on perceptual guidelines. It then maps concepts in the taxonomy to visual channels to create glyphs, with higher-level concepts mapped to channels that pop out more. This allows creating scalable glyphs for thousands of concepts to aid exploration and comparison of experimental workflows.
IMPORTANT: If you want to get a clear review of the Differences & Complementarities Between « Heuristic » and « Mathematical » approaches, we invite you to download our presentation given during the EPA (European Psychiatric Association) conference in 2011 that is now utilized in training programs.
Main single agent machine learning algorithmsbutest
This document summarizes several machine learning algorithms and their potential applications to multi-agent systems. It describes algorithms such as decision trees, neural networks, Bayesian methods, reinforcement learning, inductive logic programming, case-based reasoning, support vector machines, and genetic algorithms. For each algorithm, it provides a brief description and discusses any existing or potential work applying the algorithm to multi-agent domains.
Substructrual surrogates for learning decomposable classification problems: i...kknsastry
This paper presents a learning methodology based on a substructural classification model to solve decomposable classification problems. The proposed method consists of three important components: (1) a structural model that represents salient interactions between attributes for a given data, (2) a surrogate model which provides a functional approximation of the output as a function of attributes, and (3) a classification model which predicts the class for new inputs. The structural model is used to infer the functional form of the surrogate and its coefficients are estimated using linear regression methods. The classification model uses a maximally-accurate, least-complex surrogate to predict the output for given inputs. The structural model that yields an optimal classification model is searched using an iterative greedy search heuristic. Results show that the proposed method successfully detects the interacting variables in hierarchical problems, group them in linkages groups, and build maximally accurate classification models. The initial results on non-trivial hierarchical test problems indicate that the proposed method holds promise and have also shed light on several improvements to enhance the capabilities of the proposed method.
Comparison of relational and attribute-IEEE-1999-published ...butest
1. The document compares relational data mining methods to attribute-based methods for use in intelligent systems and data mining. Relational methods use first-order logic to represent background knowledge and relationships between objects, while attribute-based methods like neural networks are limited to attribute-value representations.
2. Relational methods have advantages over attribute-based methods for applications that require expressing complex logical relationships and background knowledge. They can also better handle sparse data. However, existing inductive logic programming systems for relational data mining are relatively inefficient for numerical data.
3. The paper proposes a hybrid relational data mining technique called MMDR that combines inductive logic programming with probabilistic inference. This allows it to efficiently handle
Comparison of relational and attribute-IEEE-1999-published ...
Computers overview
1. Computer Use in Medicinal
Chemistry
Overview of Computer-
1. Finding/storing information
Aided Drug Design 1. Literature searching (Medline, SciFinder…)
2. Structure searching (Protein Databank,
SciFinder)
3. Cataloging structure-activity data
2. Modeling existing lead compounds
3. Developing new lead compounds
NO
Is Target Structure Known?
YES
Modeling Existing Lead
PHARMACOPHORE-BASED APPROACHES STRUCTURE-BASED APPROACHES
Compounds
Generate Working Models of Ligands Generate Working Model of Target l QSAR
l Development of a mathematical model that describes in a predictive
manner the relationship between structure (represented by
numerical descriptors) and activity
Characterize Active Site
QSAR 3D QSAR Qualitative SAR (grid-based electrostatic potential...)
l Pharmacophore Model Development
l Finding a set of functional groups with the same geometric
arrangement in a series of compounds with a common biological
GENERATE NEW LEAD STRUCTURES activity
Propose New Lead or Optimize Existing Lead l 3D QSAR
(De Novo Design, Database Search, Combinatorial Chemistry...)
l Development of a quantitative model relating structure to biological
EVALUATE NEW STRUCTURES
activity in which the structural descriptors are values for various
properties computed at grid points in three-dimensional space
l Docking
Is Protein Structure Known?
l Development of a model complex of a biological target and a ligand
NO YES
l Free Energy Perturbation
QSAR or 3D QSAR model, Docking, FEP, Hydration Free Energy, l A computational method to determine the differences in free energy
Hydration Free Energy... Regression Methods... involved in transferring different ligands from the aqueous solution to
a binding site in a biological target
Synthesize/Test Best Candidates
Group Discussion Group Discussion Points
l Questions
l Identify some important questions or l QSAR – Can QSAR be used with other identification
limitations of technique based on concepts processes? (spectroscopic)
l Pharmacophore Modeling – Need to determine
from organic chemistry pharmacophore grps in each molecule with similar
characteristics
Typical chapter titles in organic chemistry textbooks: l Docking – need structures (stereochemistry often not
known for initial lead compounds)
Structure and bonding; Bonding and molecular properties; Alkanes and
cycloalkanes; Stereochemistry; Overview of organic reactions; Alkenes; Alkynes; l Limitations
Alkyl halides; Nucleophilic substitutions and eliminations; Structure determination l QSAR – No visual aspect (how to improve activity not
(spectroscopy); Conjugated dienes; Benzene and aromaticity; Electrophilic
intuitive)
aromatic substitution; Alcohols and thiols; Ethers, epoxides and sulfides;
Nucleophilic addition to carbonyls; Carboxylic acids; Carboxylic acid derivatives; l Pharmacophore Modeling – Limited to functional groups of
Carbonyl alpha-substitution reactions; Carbonyl condensation reactions; similar charge and size
Aliphatic amines; Arylamines and phenols; Carbohydrates; Amino acids, l Docking – Does not anticipate potential chemical reactions
peptides and proteins; Lipids; Heterocycles and nucleic acids (covalent inhibition)
1
2. Pharmacophore Modeling
QSAR Example Example
l Biological activity of indoleacetic acid-like The three molecules below all target protein kinase C
synthetic hormones Each molecule can adopt a conformation with common distances separating
the circled groups
l Log(1/C) = -k1(logP)2+k2(logP)+K3σ+k4 O
l C: Concentration having a standard response in a Endogenous R O H OH
Activator R O OH N CH3
standard time R Antitumor
O O Compound
l P: Octanol/water partition coefficient (S)-DAG
O OH O OCH3
l Log P reflects pharmacokinetic influence on activity –
O O
does the compound get where it needs to go?
σ reflects pharmacodynamic influence on activity – R O O R HO
l O
does the electronic nature of the compound induce H3C CH3 O OH O
Tumor AD 198
activity? Promoter
HO
O
H3 OH
C Phorbol Ester
O
3D QSAR Example
J Mol. Graph. Model. 21 (2003) 263-272
Docking Example
Blue: Negative charge disfavored Docking was used to identify the
binding site of a phospholipid in a G
Red: Negative charge favored
protein-coupled receptor
Three key ion pairing interactions
between the receptor and the
phospholipid are highlighted in panel
C
Experimental mutation of ARG120,
GLU121, and ARG292 to ALA resulted
in complete loss of phospholipid
Green: Sterically disfavored binding
Yellow: Sterically allowed
Analysis Exercise Free Energy Perturbation
l Visually examine the 1HNI structure of HIV
reverse transcriptase ∆GBind1
Ligand 1 Solvated Ligand 1 Bound
l Focus on the inhibitor and the surrounding ∆GSolv ∆GInter
∆GBind2
residues Ligand 2 Solvated Ligand 2 Bound
l What type of intermolecular interactions can
you identify visually? Most useful quantity to compare drug candidates:
∆GBind1 – ∆GBind2
l Which ones do you think are most important? Most computationally feasible quantity:
∆GSolv – ∆GInter
Since free energy is a state function, any path with the same beginning and
end points has the same value, therefore ∆GBind1+ ∆GInter = ∆GSolv + ∆GBind2
Rearrangement demonstrates the previous differences are equivalent
2
3. Is Target Structure Known?
NO YES
Developing New Leads PHARMACOPHORE-BASED APPROACHES STRUCTURE-BASED APPROACHES
Generate Working Models of Ligands Generate Working Model of Target
l De novo Design
l Techniques that build a potential ligand into the
Characterize Active Site
environment of a biological target of known structure QSAR 3D QSAR Qualitative SAR (grid-based electrostatic potential...)
l Database searching GENERATE NEW LEAD STRUCTURES
l Use of pharmacophore models to query a database for Propose New Lead or Optimize Existing Lead
new structures that also contain the requisite 3D (De Novo Design, Database Search, Combinatorial Chemistry...)
arrangement of functional groups EVALUATE NEW STRUCTURES
l Combinatorial library design
Is Protein Structure Known?
l Use of computers to determine a library of compounds NO YES
enriched in potentially active compounds that can be
QSAR or 3D QSAR model, Docking, FEP, Hydration Free Energy,
synthesized combinatorially and rapidly screened Hydration Free Energy... Regression Methods...
Synthesize/Test Best Candidates
Reading Assignment
l The Organic Chemistry of Drug Design and
Drug Action
l Chapter 2:
l Section 2.2 A, C, D, G1, H, I
l Textbook of Drug Design and Discovery
l Sections 4.1-4.3
l Sections 5.1-5.2
3