Big Data Analysis and Application of Liver Cancer Gene Sequence Based on Second-Generation Sequencing Technology

Area of Science:

• Bioinformatics
• Computational Biology
• Data Science

Background:

• The exponential growth of big data, particularly omics data (mRNA, microRNA, DNA methylation), presents challenges for traditional analysis methods.
• Existing clustering algorithms often struggle with high-dimensional data due to attribute independence assumptions, limiting classification effectiveness.

Purpose of the Study:

• To develop and evaluate a novel grid coding approach for dimensionality reduction and improved classification of big omics data.
• To compare the performance of traditional clustering methods and identify the optimal classification model for identifying therapeutic targets.

Main Methods:

• Applied traditional clustering methods (kmeans, K-nearest neighbors, hierarchical clustering, affinity propagation, nonnegative matrix decomposition) to omics data.
• Developed a multilevel grid coding strategy to map high-dimensional data to a one-dimensional representation, simplifying complexity and improving efficiency.
• Handled nondiscrete and missing values within the dataset.
• Compared classification performance across different models, including Naive Bayes, using iterative expectation-maximization steps and mass spectrometry (MS) for identification.

Main Results:

• The proposed grid coding method simplifies data complexity by converting two-dimensional data to a one-dimensional binary representation, enhancing data organization and storage.
• This approach maintains the original spatial data organization and provides stable classification efficiency, overcoming limitations of attribute independence assumptions.
• The method effectively processes nondiscrete and missing values, leading to improved classification outcomes.
• Naive Bayes emerged as the optimal classification model among those compared.

Conclusions:

• The novel grid coding technique offers a robust solution for dimensionality reduction and enhanced classification in big omics data analysis.
• This method facilitates the identification of protein targets for small molecule therapy by improving data processing and classification accuracy.
• Naive Bayes demonstrates superior performance for this classification task, paving the way for more effective therapeutic target identification.