Bioinformatics - Code of Gene-iuses

Event Overview: It is an exciting, interdisciplinary hackathon open to students of all years. The event combines the worlds of biology, coding, and machine learning in a unique attack-and-defense competition format. Participants will face rapid quizzes, hands-on bioinformatics challenges, and machine learning problem statements that push them to think creatively under time pressure. More than just solving problems, teams will strategize, defend, and “attack” with their solutions—bringing the thrill of competition to technical learning. This fusion of bioinformatics and ML creates an unforgettable experience that sharpens problem-solving, teamwork, and innovation while making learning genuinely exciting. Event Structure: Round 1: Online Quiz Description: MCQ-based quiz testing biology, maths, stats, and CS basics. Requirements: Internet-enabled device, Unstop platform access. Evaluation: Auto-graded, top teams qualify. Note: Shortlisted teams will be notified through emails. Timeline: Start: 5th October, 2025 – 10:00 AM End: 5th October, 2025 – 10:00 PM Round 2: Live Challenge Battle (Offline) Description: Teams compete on organizer + peer-broadcasted problems with real-time leaderboard. Requirements: Laptop/PC, stable internet, competition portal. Evaluation: Points for correct solves, penalties for unsolved broadcasts. Top 4–5 advance. Timeline: Start: 10th October, 2025 – 10:00 AM End: 10th October, 2025 – 1:00 PM Final Round: ML in Bioinformatics (Offline) Description: Teams implement ML models on real bioinformatics datasets, then present to judges. Requirements: Laptops, dataset access, coding environment. Evaluation: Model Performance – 40% Methodology & Innovation – 25% Reproducibility & Code Quality – 15% Presentation – 20% Timeline: Start: 10th October, 2025 – 8:00 PM End: 11th October, 2025 – 8:00 AM Detailed Round Descriptions: Round 2: Live Challenge Battle (Offline) Format: Teams solve organizer-given and peer-broadcasted problems in real time on the competition portal. Gameplay: Each team starts with base points. Correct solves earn points; incorrect or unsolved broadcasts may carry penalties. Teams can “attack” by broadcasting problems and “defend” by solving incoming ones. Deliverables: Solutions submitted on the portal (auto-graded where possible). For broadcast problems: statement + expected answer + brief verification notes. Evaluation: Organizer Questions: +10 points per correct solve. Broadcasted Questions: +25 points per correct solve. Penalty: −5 if an external broadcast remains unsolved. Activity Bonus: +1 per near-correct attempt (max +5). Leaderboard ranks updated live. Advancement: Top 4–5 teams move to the Final Round. Special Recognition: Teams that broadcast creative, high-quality ML challenges (even if not finalists) will be rewarded with goodies and swags. Basic Example Problem Statements For Round 2: Create a synthetic DNA sequence dataset with missing base-pairs and noise. Teams must clean and reconstruct the dataset to restore sequence accuracy. Provide adversarial protein sequence data with swapped labels. Teams must detect anomalies and relabel correctly using rules/ML techniques. Round 3: ML in Bioinformatics (Offline Finals): Format: Teams tackle real-world ML problems in bioinformatics such as gene clustering, protein classification, or cancer prediction. Deliverables: Code notebook/script + trained model outputs. PPT summarizing methodology, results, and insights. Visualizations (e.g., confusion matrices, clustering plots). Presentation: Each team presents for 10–15 minutes before judges, explaining their approach, innovations, and findings. Evaluation Criteria: Model Performance – 40% Methodology & Innovation – 25% Reproducibility & Code Quality – 15% Presentation & Communication – 20% Basic Example Problem Statements For Round 3: Data Clustering on Gene Expression Data Problem: High-dimensional gene expression datasets are complex and noisy, making it difficult to identify meaningful groups of genes or patients. Traditional analysis struggles to uncover hidden biological patterns. Challenge: Perform data clustering on a given gene expression dataset to identify natural clusters among the patients, visualise the results and interpret the biological significance of clusters. Protein Classification Challenge Problem: Understanding protein function from sequence and physicochemical properties is a central challenge in bioinformatics. Given a synthetic protein dataset with amino acid sequences, calculated properties, and functional classes, teams must build a machine learning model to classify proteins into their respective functional categories. Challenge: Perform exploratory data analysis (EDA) on protein sequences and their physicochemical features. Train at least two machine learning models (e.g., Random Forest, SVM, Logistic Regression, Neural Networks). Compare their performance on the test set. Use feature importance/analysis to explain which properties drive classification. Present a confusion matrix and classification metrics (accuracy, precision, recall, F1-score). Evaluation Rules & Guidelines: Winners selected based on final round judging criteria. Disqualification: plagiarism, rule violation, or misconduct. Selection for next round based on leaderboard/top scores. Eligibility Criteria & Participation Details: Open to all students from any discipline in India. Teams must have 2–4 members. +10 bonus points if the team has at least one female member. Cross-college/cross-specialization teams allowed. A student can join only one team. Submissions after the deadline will not be considered. Hackathon Requirements (Final Round Submissions): GitHub Repository: Must contain all code, scripts, and documentation. README encouraged with setup and usage instructions. PowerPoint Presentation (PPT): Problem, solution, key features, and results. Include screenshots/visualizations of ML results. General Rules & Guidelines: Organizers hold full authority on rules and decorum. Participants must follow the supervision team’s instructions. Plagiarism or invalid broadcasts = disqualification. Judges’ decisions are final. Note: Examples of problem statements shared are only for reference. The actual level of difficulty and final problem statements will be disclosed on the event day.