Publications
Primer C-VAE: An Interpretable Deep Learning Primer Design Method to Detect Emerging Virus Variants (2025)
Wang, Hanyu and Tsinda, Emmanuel K and Dunn, Anthony J and Chikweto, Francis and Zemkoho, Alain B. arXiv preprint arXiv:2503.01459.
Projects
🧬 Primer C-VAE: An Interpretable Deep Learning Primer Design Method to Detect Emerging Virus Variants (2022 / 2025)
CORMSIS External Summer Project (Master Graduation Project)
supervised by Dr. Alain B. Zemkoho and Dr. Emmanuel Kagning-Tsinda.
This project employs convolutional neural networks and variational autoencoders to identify viral variants and extract genomic features for optimal primer design. The system achieves over 98% accuracy in classifying SARS-CoV-2 variants and generates forward and reverse primers validated through in silico PCR. Beyond ~30 kb single-stranded positive-sense RNA viruses, the method successfully handles larger genomes including E. coli and S. flexneri (4.6–5 Mb), offering significant advantages over traditional tools like Primer3Plus, which is limited to 3,000 bp sequences and struggles with highly similar organisms.
Initially completed in 2022 (arXiv:2209.13591). Updated: March 2025 (arXiv:2503.01459)
🏃♂️ Motion Modal Recognition Based on Machine Learning Methods (2020)
Undergraduate graduation project (Southwest Jiaotong University)
supervised by Dr. Hua Meng (孟华).
Addressing the growing demand for motion recognition in wearable devices, this project developed an end-to-end machine learning-based motion modal recognition system. Dual inertial measurement units (IMUs) were deployed to synchronously collect wrist and foot data, followed by preprocessing (noise reduction, filtering, frequency-domain transformation) and feature engineering to enable both individual identification and motion pattern classification across 8 human motions (walking, running, cycling, etc.). The system incorporates LSTM-based anomaly data reconstruction to enhance robustness and data quality. This comprehensive solution spans data acquisition, preprocessing, feature extraction, model recognition, and anomaly repair, demonstrating high accuracy and practical applicability for daily monitoring, sports fitness, and medical rehabilitation.
🚕 A Comparative Study on the Behavior of Traditional Taxi and For-Hire Vehicle Based on Big Data (2019)
University SRTP (Student Research Training Program) at Southwest Jiaotong University
supervised by Dr. Hongtai Yang (杨鸿泰).
Based on NYC official TLC (Taxi and Limousine Commission) trip data, this project conducts a comparative analysis of spatiotemporal evolution, regional preferences, and travel patterns between traditional taxis (Yellow/Green Cabs) and for-hire vehicles (FHV). Utilizing R and Python for data mining and visualization, the study examines service distribution changes, demand-supply dynamics, and user behavior differences across NYC boroughs. The analysis provides insights into potential policy implications and market trends in urban transportation, offering evidence-based perspectives on the competitive landscape between traditional taxi services and ride-sharing platforms.
🧪 Research on the Recognition and Function of the Enhancers in HMEC and MCF-7 Cells (2018)
Provincial SRTP (Student Research Training Program) at Southwest Jiaotong University
supervised by Dr. Zhiyun Guo (郭志云).
This project focuses on the transcriptional regulatory mechanisms of breast cancer, conducting systematic research around enhancers, transcription factors, and target genes. Given China's heavy burden of annual new cancer cases and deaths, with breast cancer being one of the most common cancers among women, this study aims to analyze key regulatory factors and pathways in breast cancer from the transcriptional regulatory level, providing a theoretical basis for subsequent intervention strategies. The research employs DNase-seq and histone modification data (H3K27ac, H3K4me1, H3K4me3) to identify genomic enhancers in mammary epithelial and breast cancer cell lines.
View all my projects on GitHub.