20–21 Nov 2026
Online Conference
Asia/Karachi timezone

GeneNet: Gene Co-Expression Network Analysis and Ensemble Learning for Multiclass Cancer Tissue Classification

21 Nov 2026, 13:45
5m
Online Conference

Online Conference

Online Conference

Speaker

Mr Mudassir Muneer

Result

The pipeline reduced the feature set from 20,531 genes to 500 and generated a co-expression graph with 500 vertices, 1,814 edges, and 304 connected components. Project presentation results reported 99.75% classification accuracy, 99.86% precision, 99.72% recall, and 3.87% out-of-bag error. These figures are preliminary: the original evaluation protocol, confusion matrix, and raw test logs were unavailable for independent verification. Clinical diagnostic validity was not established.

Background

RNA sequencing provides valuable information
about cancer biology but produces high-dimensional datasets that are challenging to analyze efficiently. Identifying informative genes and their relationships may support computational classification of cancer tissue types. Existing approaches often rely on standard machine-learning libraries, while transparent, educational implementations combining gene co-expression networks with custom ensemble classifiers remain valuable for demonstrating interpretable data analysis.

Discussion (methods + interpretation)

This project aimed to develop GeneNet, a C++ pipeline that reduces high-dimensional RNA-Seq data, explores gene co-expression relationships, and classifies samples into five cancer tissue types. The work demonstrates how custom data structures, graph algorithms, and ensemble learning can be integrated for computational cancer research.

Aim & Purpose

This project aimed to develop GeneNet, a C++ pipeline that reduces high-dimensional RNA-Seq data, explores gene co-expression relationships, and classifies samples into five cancer tissue types. The work demonstrates how custom data structures, graph algorithms, and ensemble learning can be integrated for computational cancer research.

Email address 24G-BCS020@agu.edu.pk
Phone number 03355818818

Authors

Presentation materials