About Workshop
This hands-on workshop introduces participants to AI-driven nanomaterials discovery using materials databases, materials informatics, machine learning and explainable AI. Participants will learn how to retrieve and process materials data, generate composition- and structure-based descriptors, develop predictive models for key material properties, computationally screen candidate materials, and use explainable AI to understand which physicochemical features control material performance.
Aim
To provide participants with practical skills in materials informatics and artificial intelligence for accelerating nanomaterials discovery, enabling them to move from materials databases and descriptors to property prediction, candidate screening and interpretable materials-design decisions.
What Participants Will Learn
- Understand the role of AI and materials informatics in modern materials discovery.
- Explore databases containing computational and experimental materials information.
- Retrieve and organise material composition, structural and property data.
- Use Python-based materials-science libraries for data preparation.
- Generate meaningful composition- and structure-based materials descriptors.
- Perform exploratory analysis of structure–property relationships.
- Build machine-learning models for materials-property prediction.
- Compare regression and classification approaches for materials screening.
- Apply Random Forest and XGBoost to materials datasets.
- Evaluate models using appropriate performance metrics.
- Use explainable AI to identify important material descriptors.
- Perform computational screening of candidate materials.
- Rank promising materials based on predicted properties.
- Interpret AI predictions from a materials-science perspective.
- Develop a reproducible workflow for AI-guided materials discovery.
Structure
Day 1: Disease Variant Interpretation & Prime-Editing Strategy
Core Objective: Identify a clinically relevant disease-associated variant, interpret its genomic and transcript context, and translate it into a suitable input for prime-editing correction design.- Introduce precision genome engineering and compare prime editing with conventional CRISPR-Cas editing and base-editing approaches.
- Identify disease-associated variants using ClinVar, interpret genomic coordinates, alleles, transcripts and variant annotations, and assess functional consequences using Ensembl VEP.
- Retrieve reference genomic and transcript sequences, examine the local sequence context, and determine suitable preliminary correction strategies including PE2, PE3, PE3b and twinPE approaches.
Hands-on Lab: Select a clinically relevant variant from ClinVar, identify the associated gene and transcript, annotate its predicted consequences using Ensembl VEP, explore the genomic locus in the UCSC Genome Browser, and prepare the reference and desired-edited sequences required for downstream prime-editing design.
Tools Covered: ClinVar, Ensembl, Ensembl VEP, UCSC Genome Browser, NCBI, curated variant datasets
Output: Annotated disease variant, validated reference sequence context and prime-editing-ready variant correction input.
Day 2: AI-Guided pegRNA Design & Prime-Editing Outcome Prediction
Core Objective: Generate candidate prime-editing guide designs and use AI-based prediction to compare expected editing efficiencies and prioritise promising editing strategies.- Understand pegRNA architecture, including spacer, primer-binding site (PBS) and reverse-transcription template (RTT), and examine how sequence context influences prime-editing efficiency.
- Generate computational pegRNA and nicking-guide designs for PE2, PE3 and PE3b strategies, and explore twinPE approaches for more complex sequence modifications where applicable.
- Apply mechanistic machine-learning concepts and OptiPrime-based prediction to compare candidate designs according to sequence features and predicted editing outcomes.
Hands-on Lab: Generate candidate pegRNA designs for the selected disease variant, compare spacer, PBS and RTT configurations, evaluate PE3/PE3b or twinPE alternatives where relevant, and use OptiPrime predictions to construct a comparative table of expected prime-editing outcomes.
Tools Covered: OptiPrime, PrimeDesign, Python, Google Colab, sequence-analysis tools
Output: Candidate pegRNA and nicking-guide designs with AI-predicted editing efficiencies and a comparative prime-editing design table.
Day 3: Prime-Editing Optimisation, Candidate Ranking & Variant-Correction Design
Core Objective: Integrate predicted editing efficiency, sequence-design quality, unintended-edit considerations and biological relevance to prioritise the most suitable prime-editing correction strategy.- Compare multiple pegRNA candidates using predicted editing efficiency, sequence-context features, design-quality criteria and expected editing-outcome profiles.
- Evaluate desired versus unintended editing outcomes, PE3 and twinPE strategy differences, mismatch-repair-related considerations and other factors that may influence prime-editing performance.
- Apply multi-parameter candidate prioritisation, interpret AI prediction confidence and limitations, and translate computational predictions into a research-oriented therapeutic variant-correction strategy.
Hands-on Lab: Compare predicted performance across multiple prime-editing candidates, rank designs using efficiency and design-quality criteria, examine potential unwanted-edit considerations, compare alternative correction strategies, and prepare a final Prime-Editing Variant Correction Design Report.
Tools Covered: OptiPrime, PrimeDesign, Python, Google Colab, Pandas, candidate-ranking templates
Output: Ranked prime-editing candidates, comparative outcome analysis and a final evidence-based Prime-Editing Variant Correction Design Report.Important Dates
Registration Ends
4: 30 PM
Workshop Dates
2026-08-31
5:30 PM
5:30 PM
What You Will Gain
Outcomes
- Convert chemical composition and crystal-structure information into machine-learning-ready datasets.
- Generate composition- and structure-based descriptors using pymatgen and matminer.
- Build and evaluate AI/ML models for materials-property prediction.
- Apply Random Forest and XGBoost for regression, classification and candidate screening.
- Interpret model predictions using SHAP and feature-importance analysis.
- Identify key physicochemical and structural features influencing material properties.
- Perform high-throughput computational screening of candidate materials.
- Rank promising materials using predicted performance and multi-parameter criteria.
- Establish interpretable structure–property relationships using AI.
- Assess model performance, prediction confidence and limitations.
- Develop an end-to-end workflow for AI-driven materials and nanomaterials discovery.
- Generate a scientifically supported shortlist of materials for further computational or experimental validation.
- Prepare a final AI-Guided Materials Discovery & Candidate Screening Report.
Who Should Attend
- Nanoscience and Nanotechnology
- Materials Science
- Materials Engineering
- Chemistry
- Chemical Engineering
- Physics
- Applied Physics
- Metallurgical Engineering
- Ceramic Engineering
- Energy Materials
- Computational Materials Science
- Data Science and Artificial Intelligence
- Related interdisciplinary STEM disciplines