Dr. Zubaida Khatoon
Affiliate Royal Society of Biology
Lucknow, India
Actions
My online membership from BSAC endorsed my research background and enlighten my digital modelling and creativity web of open science community Amsterdam (OSCA) for research projects and grants. My Ph.D. degree certificate and a trophy become a pride indeed at the auditorium of my school of life sciences, JNU, New Delhi in the year 2007. Appreciation certificate for Life Membership from SAMA and Editorial Board Membership certificate from Edwin In, has become a passion for my research approach.
Area of Expertise
Topics
AI AGENTS AND DATA SCIENCE A SPY FOR PROTEIN MODELLING
Good Morning everyone. Thank you for being here today. My name is Dr Zubaida Khatoon, and I would like to welcome all the students, Researcher's, and Faculty to join this session to uncover the hidden architecture of life with AI Agent and now a day machine learning models have now being helping us decode the structures of life itself. In this seminar, I would like to explain through the types of AI agents being used in structural biology for protein structure and function prediction along with how they’re transforming research in protein folding, molecular design, and beyond.
In this session, I’ll walkthrough the background knowledge and types of AI agents being used in structural biology and how they’re transforming research in protein folding, molecular design, and beyond through machine learning.
Vision- pioneering a scientific domain where AI agents being a molecular detectives to unravel the mysteries of protein folding and enabling the design of life-saving therapeutics with unprecedented speed and accuracy.In this session, my visionary intent has been to co-create a roadmap with all challenges and conventional thinking towards AI Agent programming and application. My mission and perspective will focus on pioneering a scientific domain where AI agents being a molecular detectives to unravel the mysteries of protein folding and enabling the design of life-saving therapeutics with unprecedented speed and accuracy.
Roadmap
I would like to start with a quick overview of AI agent types, further, discussion for specific use of AI Agents in structural biology, and At the last Discuss about future perspectives and open challenges in the domain of protein modelling. AI Agent being the choice of real people in the real world become an amazing tool to help in real life by just performing simple tasks in a moment of time otherwise it requires a lot of work with high effort for long time. An AI agent can have a significant impact on personalized learning through the development of tools, software, and pipelines designed to be embedded into educational systems. It provides immense knowledge about structured information to display on the screen and having broad impact and transformative potential of AI across various sectors, for example it’s immense application and impact on AI-assisted project management, innovative problem-solving sessions, enhanced environmental initiatives, interactive historical learning experiences etc. An AI agent can emerge through the pipeline of Data Repository server for example a pipeline connected to protein structure prediction, drug discovery, and molecular dynamics all three linked to PDB and PDBe. Data mining software’s provide plenty of data to align meaningful information to represent in graphical patterns for calligraphy, knowledge and statistical calculations etc. AI agents an autonomous software programs that use machine learning algorithms to learn from previous existing data and design a changing scenarios for improvement and adaptation. AI agents can streamline operations for an improved efficiency reduced the costs and boosts the productivity. AI agents through Zapier provide automation by use of chatbots to connect patients with medical professionals abroad in real-time. The ProtChat being an AI agent enables researchers to command directly for an improve efficiency and usability significantly and make suitable without a computational background. An AI agent designed as a compound system having modules and each implementing a distinct functionality. These modules being focusing on perception, interaction, memory, and reasoning. For example through a hypothetical AI Agent, the researchers can combine the perception, interaction, memory, and reasoning modules to study the selection against pathogenic mitochondrial DNA in the germline. Other examples of AI agent tools have been ChemCrow and WebGpt that mainly based on a large language model (LLMs). The data mining tools for science literature on web like science.gov, CORE, semantic scholar, RefSeek, scienceopen, the lens, citation gecko, local citation network, researchrabbit, open access button. The target–template alignment tools like BLAST, FASTA, NetSufp- 3.0, UNIPROT, The Swiss-PdbViewer, DOCKSURF, SCOPe, CATH, ECOD will deduce the phylogenetic tree from evolutionary homology after domain classification of protein and enzyme. The Protein designing software used in bioinformatics for structure prediction, including homology modeling, protein threading, ab initio methods, secondary structure prediction, and transmembrane helix and signal peptide prediction enlist RaptorX, Biskit, FoldX, Phyre, Phyre2, HHpred, MODELLER, CONFOLD, Molecular Operating Environment (MOE), Robetta, BHAGEERATHH, Swiss-model, Yasara, AWSEM-Suite, I-TASSER, trRosetta, ROBETTA, Rosetta@home, Abalone, C-QUARK. The AI agent in three dimensional protein designing in less time for high accuracy includes OpenProtein.AI, PoET:A, DNASTAR. These AI agent saves time from laborious lab work still have challenges on ethical framework and accuracy and data recovery tools that may be amplified with the introduction of multi-agent systems. The types of AI agents used in this domain of research known till date include:
1. Learning Agents-This AI Agent has been used for learning patterns from large biological datasets and commonly known in structural biology. They have been classified into three major categories included on-
A. Supervised Learning Agents-This AI Agent have it’s application in Prediction of protein secondary structure, contacts among residue, or in ligand binding sites. For Examples of this AI Agent include: Deep learning models pipelined on PDB data for structure-function prediction.
b. Unsupervised Learning Agents- This type of AI Agent have been used for Protein sequence clustering and in structural motif discovery. For Example: Auto encoders for embedding protein structures or sequences.
c. Reinforcement Learning Agents- They have major role in Protein folding path simulation and molecule design. For Example: AI Agents optimized for protein folding trajectories or small molecule generation. This type of AI learning Agent have been optimized to explore an reward.
2. Model-Based Agents This AI Agent have it’s application in stimulating biological systems using known physical/chemical rules along with learned models. For Example: AI-augmented force field agents for protein-ligand interaction modeling.
3. Goal-Based Agents- This agent has application in Designing, aiming to generate protein sequences that fold into target structures or bind to specific targets.Examples include-Generative Agents in inverse folding problems (sequence → structure mapping), like AlphaFold (DeepMind), RoseTTAFold (University of Washington), DeepFold
4. Multi-Agent Systems- This Agent have used to simulate protein-protein interactions, collaborative drug docking, or distributed computation of molecular dynamics. Examples include- Agents simulating cellular environments or distributed simulations of molecular interactions. Some example for simulation include Rosetta Framework (especially RosettaCommons), AlphaFold-Multimer (Extended from AlphaFold), AlphaFold-Multimer (Extended from AlphaFold), Foldit (Crowdsourced Human-AI System), BioGrid or Grid-Based Simulations (e.g., Folding@home).
5. Cognitive or Expert Agents- This Agent has its use in Assistance of researchers in hypothesis generation, literature review, or experimental design. AI copilots trained on structural biology literature (e.g., GPT-based tools fine-tuned on PubMed). ProFunc, AlphaFold (especially AlphaFold 2), SWISS-MODEL Expert System, DeepMind’s Protein Language Models (e.g., Enformer for DNA, related to AlphaFold’s embeddings), FoldX, RaptorX being an example.
6. Generative Agents (a subtype of learning agents)- This type of AI Agent has its own use in De novo protein design, structure generation. We can enlist the Examples being given below-
(i) AlphaFold (DeepMind): Predicts 3D structure from sequence.
(ii) RoseTTAFold (Baker Lab): Similar predictive generative model.
(iii) ESMFold: Uses large language models for structure prediction.
(iv) ProteinMPNN (David Baker Lab)
(v) RFdiffusion (RoseTTAFold Diffusion)
(vi) ProtGPT2
(vii) ProGen (Salesforce Research)
(viii) Chroma (Generate Biomedicines)
7. Conversational or Interactive Agents- This AI Agent have Interfaces for exploring biological databases or interpreting AI model outputs. Example include: Assistants that help biologists query protein structure databases in natural language. Foldit (University of Washington), ChatGPT and Protein Modeling Plugins or Custom Tools, DeepChain™ (by InstaDeep), IBM Watson for Drug Discovery (now deprecated but conceptually relevant), BioBERT / SciBERT and Chat Wrappers, Interactive Jupyter-based Platforms (e.g., ColabFold).
AI Agent programming and protein folding represent two of the most transformative areas in computer science and biotechnology. Though seemingly distinct, their intersection is revolutionizing drug discovery, disease understanding, and computational biology.AI Agent Programming involves the development of autonomous systems— An Agent— that ascertain to their environment, having right approach enact to set and bounce specific goals. The agent have been governed by algorithms having range from simple rule-based systems to complex machine learning models. Advanced AI agents updated on chronological data to adapt their strategy of learning based on reinforcement, deep learning, and hybrid approach and become invaluable in solving complex real-world problems. We can design an AI Agent with background knowledge, which includes introductory information about a person's profession, expertise, and premises. After programming with background knowledge log into ChatGPT and dumped into it. The background knowledge should be in csv format and include the following information.
Name: Jules White.
Profession: Professor in Computer Science at Vanderbilt University and Senior Advisor to the chancellor in Generative AI Enterprise and Education Solutions.
Premises: Nashville, Tennessee,
Firstly, write program on HTML using Tailwind CSS (Cascading Style Sheets) with background knowledge in CSV and highlight everything on Vanderbilt's webpage here, and proceeds to dumped it into ChatGPT. It highlights the used cases, sample prompts, and upcoming AI tools, tailored to your work. Programming is all about telling it what's going to happen in the future. No matter what you give it, it turns it into data. It has impact on various application included on. In building an office applications, in working applications, and in all kinds of complex things from simple things. Now we're going to talk about tools with GPTs. AI Agent is an Agentic AI that manifest in a GPT. It can actually go and access tools and use sets of tools to where it has been pooled via pipeline. In this case, we have two agents that have been checked-
1. Web browsing - GPT with Browsing, Research Assistant Agents for example PubMed or arXiv, Autonomous Agents (e.g., AutoGPT, BabyAGI).
2. Generative AI- GPT-4 or GPT-3.5 (OpenAI), Claude (Anthropic), Gemini (Google DeepMind), Multimodal generative AI combining text, image, and code generation, LLaMA, Mistral, ProGen or ProtGPT2, GraphGPT or Molecule Chef or ChemCrow. Codex or Copilot or CodeWhisperer,
Protein Folding, on the other hand, is a biological process in which a polypeptide chain folds into a specific three-dimensional structure, which determines its function in the body. In other words, it depicts the three-dimensional geometry and spatial arrangement of the atom in an amino acid molecule. Misfolded proteins have been linked to numerous diseases, including Alzheimer's, Parkinson's, and cystic fibrosis. Understanding and predicting protein folding has been a long-standing challenge in biology due to the astronomical number of possible configurations a protein could theoretically adopt. Understanding and predicting of protein folding has been a long-standing challenge in biochemistry due to the astronomical number of possible configurations a protein could theoretically adopt. The convergence of AI agent programming and protein folding and design passed over the landmark of DeepMind's AlphaFold. It is an AI system having capability of protein structure prediction having near-experimental accuracy. AlphaFold being an essential AI agent has been trained on vast datasets of known protein structures and sequences. It demonstrates how AI agents can learn complex physical rules and patterns without comprehensive programming, reckoning on statistical learning and optimization techniques. This implement innovation to exemplify the power of AI agent programming to tackle problems of enormous complexity. It opens avenues for applications such as rapid vaccine development, Drug discovery, personalized medicine, and synthetic biology. AI agents being now become useful in structure prediction, but also in designing new proteins with desired functions which has been a key goal in bioengineering.
The scientists foraged a promising computational method to predict protein folding to solve a central challenge in biology known as the “protein folding and design”. DeepMind’s AlphaFold changed the landscape by working at the advances in deep learning, further-on AlphaFold achieved a pioneering in protein prediction accuracy, that competes with experimental methods. This advancement through AI and machine learning approaching towards a spacious by-products in protein engineering and drug design. Deep Mind Developed AlphaFold 1 and 2 during 2018 and 2020 for prediction of the three-dimensional structure of a protein from its amino acid sequence. The observance of statistical patterns of protein folding of known structures in the Protein Data Bank (PDB) and by phylogenetic comparison of sequence embedded in multiple sequence alignments. AlphaFold2, used a novel architecture based on transformer networks attention mechanisms. On combining both spatial and evolutionary information, it produces highly accurate models of protein structures. AlphaFold2 achieve near-experimental accuracy with the Critical Assessment of protein Structure Prediction (CASP14) competency. AlphaFold’s models allow scientists to study proteins whose structures were previously unknown, speeding up discoveries in cell biology, neurobiology, and disease mechanisms. Moreover, With structural insights into proteins related to diseases like cancer, Alzheimer’s, and COVID-19, AlphaFold enables more targeted drug design and protein-protein interaction studies. Further, DeepMind and EMBL-EBI developed the AlphaFold Protein Structure Database. This database hosts predicted structures for hundreds of millions of proteins from nearly all known organisms. Scientists use AlphaFold to design new proteins with known function of interest. The AlphaFold, CASP, ECOD and SCOPe help In completing the innovative projects in biotechnology, green chemistry, and therapeutics.
Challenges:
Despite its strengths, AlphaFold is not perfect and it being a challenge in dynamics of proteins and complexes where it have struggling for proteins that adopt multiple conformations or interact in large complexes. In membrane proteins and intrinsically disordered regions it remains difficult for right conformation prediction due to their complexity or lack of defined structure. No direct prediction of function by AlphaFold whether it can precisely predicts structure, but have limitations in depicting how the protein behaves in vivo.
Conclusion:
AlphaFold represents a landmark achievement in artificial intelligence and biology. It not only helpful in long-standing scientific problem but also provide valuable information to AI driven transformative breakthroughs in the life sciences. Through the development of science with tools like AlphaFold-Multimer and RoseTTAFold the scope of structural biology will begin towards automation, and open avenues for global collaboration. The integration of AI agent programming with protein folding and design depicts the profound impact of computational intelligence on biological research. An AI Agents development continuing for its role in decoding life’s fundamental processes for growth, leading to scientific advances with far-reaching implications for health, industry, and for our understanding of life itself.
I would like to request to please click on the following link to catch the favourite session-
https://www.veed.io/edit/1777fbd5-c92f-4952-bf28-fbbe40004e63
https://www.veed.io/view/1777fbd5-c92f-4952-bf28-fbbe40004e63?panel=share
AI AGENTS AND DATA SCIENCE A SPY FOR PROTEIN MODELLING
Good Morning everyone. Thank you for being here today. My name is Dr Zubaida Khatoon, and I would like to welcome all the students, Researcher's, and Faculty to join this session to uncover the hidden architecture of life with AI Agent and now a day machine learning models have now being helping us decode the structures of life itself. In this seminar, I would like to explain through the types of AI agents being used in structural biology for protein structure and function prediction along with how they’re transforming research in protein folding, molecular design, and beyond. The goal of this session
has been to provide background knowledge of AI Agents including types of AI agents being used in structural biology and how they’re transforming research in protein folding, molecular design, and beyond through machine learning.
In this session, my visionary intent has been to co-create a roadmap with all challenges and conventional thinking towards AI Agent programming and application. My mission and perspective will focus on pioneering a scientific domain where AI agents being a molecular detectives to unravel the mysteries of protein folding and enabling the design of life-saving therapeutics with unprecedented speed and accuracy.
Roadmap
I would like to start with a quick overview of AI agent types, further, discussion for specific use of AI Agents in structural biology, and At the last Discuss about future perspectives and open challenges in the domain of protein modelling. AI Agent being the choice of real people in the real world become an amazing tool to help in real life by just performing simple tasks in a moment of time otherwise it requires a lot of work with high effort for long time. An AI agent can have a significant impact on personalized learning through the development of tools, software, and pipelines designed to be embedded into educational systems. It provides immense knowledge about structured information to display on the screen and having broad impact and transformative potential of AI across various sectors, for example it’s immense application and impact on AI-assisted project management, innovative problem-solving sessions, enhanced environmental initiatives, interactive historical learning experiences etc. An AI agent can emerge through the pipeline of Data Repository server for example a pipeline connected to protein structure prediction, drug discovery, and molecular dynamics all three linked to PDB and PDBe. Data mining software’s provide plenty of data to align meaningful information to represent in graphical patterns for calligraphy, knowledge and statistical calculations etc. AI agents an autonomous software programs that use machine learning algorithms to learn from previous existing data and design a changing scenarios for improvement and adaptation. AI agents can streamline operations for an improved efficiency reduced the costs and boosts the productivity. AI agents through Zapier provide automation by use of chatbots to connect patients with medical professionals abroad in real-time. The ProtChat being an AI agent enables researchers to command directly for an improve efficiency and usability significantly and make suitable without a computational background. An AI agent designed as a compound system having modules and each implementing a distinct functionality. These modules being focusing on perception, interaction, memory, and reasoning. For example through a hypothetical AI Agent, the researchers can combine the perception, interaction, memory, and reasoning modules to study the selection against pathogenic mitochondrial DNA in the germline. Other examples of AI agent tools have been ChemCrow and WebGpt that mainly based on a large language model (LLMs). The data mining tools for science literature on web like science.gov, CORE, semantic scholar, RefSeek, scienceopen, the lens, citation gecko, local citation network, researchrabbit, open access button. The target–template alignment tools like BLAST, FASTA, NetSufp- 3.0, UNIPROT, The Swiss-PdbViewer, DOCKSURF, SCOPe, CATH, ECOD will deduce the phylogenetic tree from evolutionary homology after domain classification of protein and enzyme. The Protein designing software used in bioinformatics for structure prediction, including homology modeling, protein threading, ab initio methods, secondary structure prediction, and transmembrane helix and signal peptide prediction enlist RaptorX, Biskit, FoldX, Phyre, Phyre2, HHpred, MODELLER, CONFOLD, Molecular Operating Environment (MOE), Robetta, BHAGEERATHH, Swiss-model, Yasara, AWSEM-Suite, I-TASSER, trRosetta, ROBETTA, Rosetta@home, Abalone, C-QUARK. The AI agent in three dimensional protein designing in less time for high accuracy includes OpenProtein.AI, PoET:A, DNASTAR. These AI agent saves time from laborious lab work still have challenges on ethical framework and accuracy and data recovery tools that may be amplified with the introduction of multi-agent systems. The types of AI agents used in this domain of research known till date include:
1. Learning Agents-This AI Agent has been used for learning patterns from large biological datasets and commonly known in structural biology. They have been classified into three major categories included on-
A. Supervised Learning Agents-This AI Agent have it’s application in Prediction of protein secondary structure, contacts among residue, or in ligand binding sites. For Examples of this AI Agent include: Deep learning models pipelined on PDB data for structure-function prediction.
b. Unsupervised Learning Agents- This type of AI Agent have been used for Protein sequence clustering and in structural motif discovery. For Example: Auto encoders for embedding protein structures or sequences.
c. Reinforcement Learning Agents- They have major role in Protein folding path simulation and molecule design. For Example: AI Agents optimized for protein folding trajectories or small molecule generation. This type of AI learning Agent have been optimized to explore an reward.
2. Model-Based Agents This AI Agent have it’s application in stimulating biological systems using known physical/chemical rules along with learned models. For Example: AI-augmented force field agents for protein-ligand interaction modeling.
3. Goal-Based Agents- This agent has application in Designing, aiming to generate protein sequences that fold into target structures or bind to specific targets.Examples include-Generative Agents in inverse folding problems (sequence → structure mapping), like AlphaFold (DeepMind), RoseTTAFold (University of Washington), DeepFold
4. Multi-Agent Systems- This Agent have used to simulate protein-protein interactions, collaborative drug docking, or distributed computation of molecular dynamics. Examples include- Agents simulating cellular environments or distributed simulations of molecular interactions. Some example for simulation include Rosetta Framework (especially RosettaCommons), AlphaFold-Multimer (Extended from AlphaFold), AlphaFold-Multimer (Extended from AlphaFold), Foldit (Crowdsourced Human-AI System), BioGrid or Grid-Based Simulations (e.g., Folding@home).
5. Cognitive or Expert Agents- This Agent has its use in Assistance of researchers in hypothesis generation, literature review, or experimental design. AI copilots trained on structural biology literature (e.g., GPT-based tools fine-tuned on PubMed). ProFunc, AlphaFold (especially AlphaFold 2), SWISS-MODEL Expert System, DeepMind’s Protein Language Models (e.g., Enformer for DNA, related to AlphaFold’s embeddings), FoldX, RaptorX being an example.
6. Generative Agents (a subtype of learning agents)- This type of AI Agent has its own use in De novo protein design, structure generation. We can enlist the Examples being given below-
(i) AlphaFold (DeepMind): Predicts 3D structure from sequence.
(ii) RoseTTAFold (Baker Lab): Similar predictive generative model.
(iii) ESMFold: Uses large language models for structure prediction.
(iv) ProteinMPNN (David Baker Lab)
(v) RFdiffusion (RoseTTAFold Diffusion)
(vi) ProtGPT2
(vii) ProGen (Salesforce Research)
(viii) Chroma (Generate Biomedicines)
7. Conversational or Interactive Agents- This AI Agent have Interfaces for exploring biological databases or interpreting AI model outputs. Example include: Assistants that help biologists query protein structure databases in natural language. Foldit (University of Washington), ChatGPT and Protein Modeling Plugins or Custom Tools, DeepChain™ (by InstaDeep), IBM Watson for Drug Discovery (now deprecated but conceptually relevant), BioBERT / SciBERT and Chat Wrappers, Interactive Jupyter-based Platforms (e.g., ColabFold).
AI Agent programming and protein folding represent two of the most transformative areas in computer science and biotechnology. Though seemingly distinct, their intersection is revolutionizing drug discovery, disease understanding, and computational biology.AI Agent Programming involves the development of autonomous systems— An Agent— that ascertain to their environment, having right approach enact to set and bounce specific goals. The agent have been governed by algorithms having range from simple rule-based systems to complex machine learning models. Advanced AI agents updated on chronological data to adapt their strategy of learning based on reinforcement, deep learning, and hybrid approach and become invaluable in solving complex real-world problems. We can design an AI Agent with background knowledge, which includes introductory information about a person's profession, expertise, and premises. After programming with background knowledge log into ChatGPT and dumped into it. The background knowledge should be in csv format and include the following information.
Name: Jules White.
Profession: Professor in Computer Science at Vanderbilt University and Senior Advisor to the chancellor in Generative AI Enterprise and Education Solutions.
Premises: Nashville, Tennessee,
Firstly, write program on HTML using Tailwind CSS (Cascading Style Sheets) with background knowledge in CSV and highlight everything on Vanderbilt's webpage here, and proceeds to dumped it into ChatGPT. It highlights the used cases, sample prompts, and upcoming AI tools, tailored to your work. Programming is all about telling it what's going to happen in the future. No matter what you give it, it turns it into data. It has impact on various application included on. In building an office applications, in working applications, and in all kinds of complex things from simple things. Now we're going to talk about tools with GPTs. AI Agent is an Agentic AI that manifest in a GPT. It can actually go and access tools and use sets of tools to where it has been pooled via pipeline. In this case, we have two agents that have been checked-
1. Web browsing - GPT with Browsing, Research Assistant Agents for example PubMed or arXiv, Autonomous Agents (e.g., AutoGPT, BabyAGI).
2. Generative AI- GPT-4 or GPT-3.5 (OpenAI), Claude (Anthropic), Gemini (Google DeepMind), Multimodal generative AI combining text, image, and code generation, LLaMA, Mistral, ProGen or ProtGPT2, GraphGPT or Molecule Chef or ChemCrow. Codex or Copilot or CodeWhisperer,
Protein Folding, on the other hand, is a biological process in which a polypeptide chain folds into a specific three-dimensional structure, which determines its function in the body. In other words, it depicts the three-dimensional geometry and spatial arrangement of the atom in an amino acid molecule. Misfolded proteins have been linked to numerous diseases, including Alzheimer's, Parkinson's, and cystic fibrosis. Understanding and predicting protein folding has been a long-standing challenge in biology due to the astronomical number of possible configurations a protein could theoretically adopt. Understanding and predicting of protein folding has been a long-standing challenge in biochemistry due to the astronomical number of possible configurations a protein could theoretically adopt. The convergence of AI agent programming and protein folding and design passed over the landmark of DeepMind's AlphaFold. It is an AI system having capability of protein structure prediction having near-experimental accuracy. AlphaFold being an essential AI agent has been trained on vast datasets of known protein structures and sequences. It demonstrates how AI agents can learn complex physical rules and patterns without comprehensive programming, reckoning on statistical learning and optimization techniques. This implement innovation to exemplify the power of AI agent programming to tackle problems of enormous complexity. It opens avenues for applications such as rapid vaccine development, Drug discovery, personalized medicine, and synthetic biology. AI agents being now become useful in structure prediction, but also in designing new proteins with desired functions which has been a key goal in bioengineering.
The scientists foraged a promising computational method to predict protein folding to solve a central challenge in biology known as the “protein folding and design”. DeepMind’s AlphaFold changed the landscape by working at the advances in deep learning, further-on AlphaFold achieved a pioneering in protein prediction accuracy, that competes with experimental methods. This advancement through AI and machine learning approaching towards a spacious by-products in protein engineering and drug design. Deep Mind Developed AlphaFold 1 and 2 during 2018 and 2020 for prediction of the three-dimensional structure of a protein from its amino acid sequence. The observance of statistical patterns of protein folding of known structures in the Protein Data Bank (PDB) and by phylogenetic comparison of sequence embedded in multiple sequence alignments. AlphaFold2, used a novel architecture based on transformer networks attention mechanisms. On combining both spatial and evolutionary information, it produces highly accurate models of protein structures. AlphaFold2 achieve near-experimental accuracy with the Critical Assessment of protein Structure Prediction (CASP14) competency. AlphaFold’s models allow scientists to study proteins whose structures were previously unknown, speeding up discoveries in cell biology, neurobiology, and disease mechanisms. Moreover, With structural insights into proteins related to diseases like cancer, Alzheimer’s, and COVID-19, AlphaFold enables more targeted drug design and protein-protein interaction studies. Further, DeepMind and EMBL-EBI developed the AlphaFold Protein Structure Database. This database hosts predicted structures for hundreds of millions of proteins from nearly all known organisms. Scientists use AlphaFold to design new proteins with known function of interest. The AlphaFold, CASP, ECOD and SCOPe help In completing the innovative projects in biotechnology, green chemistry, and therapeutics.
Challenges:
Despite its strengths, AlphaFold is not perfect and it being a challenge in dynamics of proteins and complexes where it have struggling for proteins that adopt multiple conformations or interact in large complexes. In membrane proteins and intrinsically disordered regions it remains difficult for right conformation prediction due to their complexity or lack of defined structure. No direct prediction of function by AlphaFold whether it can precisely predicts structure, but have limitations in depicting how the protein behaves in vivo.
Conclusion:
AlphaFold represents a landmark achievement in artificial intelligence and biology. It not only helpful in long-standing scientific problem but also provide valuable information to AI driven transformative breakthroughs in the life sciences. Through the development of science with tools like AlphaFold-Multimer and RoseTTAFold the scope of structural biology will begin towards automation, and open avenues for global collaboration. The integration of AI agent programming with protein folding and design depicts the profound impact of computational intelligence on biological research. An AI Agents development continuing for its role in decoding life’s fundamental processes for growth, leading to scientific advances with far-reaching implications for health, industry, and for our understanding of life itself.
I would like to request to click on the following given link for video clips-
https://www.veed.io/edit/1777fbd5-c92f-4952-bf28-fbbe40004e63
https://www.veed.io/view/1777fbd5-c92f-4952-bf28-fbbe40004e63?panel=share
Dr ZUBAIDA Session
My SESSION IS ON THE TOPIC ENTITLED
"AI AGENTS AND DATA SCIENCE A SPY FOR PROTEIN MODELLING.''
I would like to request to click the given link for short video clip on AI Agent-
https://www.veed.io/edit/1777fbd5-c92f-4952-bf28-fbbe40004e63
https://www.veed.io/view/1777fbd5-c92f-4952-bf28-fbbe40004e63?panel=share
Dr. Zubaida Khatoon
Affiliate Royal Society of Biology
Lucknow, India
Actions
Please note that Sessionize is not responsible for the accuracy or validity of the data provided by speakers. If you suspect this profile to be fake or spam, please let us know.
Jump to top