Missing-Values Imputation Algorithms for Microarray Gene Expression Data.

Area of Science:

• Bioinformatics
• Genomics
• Computational Biology

Background:

• Missing values are a pervasive issue in gene expression studies, impacting data interpretation and downstream analyses like cancer prediction.
• Existing bioinformatics tools for cancer prediction rely on comprehensive datasets, necessitating effective strategies for handling missing data.

Purpose of the Study:

• To review and classify existing missing-value imputation techniques for gene expression data.
• To guide scientists in selecting appropriate imputation algorithms based on data characteristics and analytical goals.

Main Methods:

• The study reviews research on missing-value imputation approaches for gene expression data.
• Algorithms are classified into global, hybrid, local, and knowledge-based techniques.
• Focus is placed on differences between algorithms using local and global data correlations.

Main Results:

• A classification framework for imputation algorithms is presented.
• The review highlights the importance of matching algorithms to specific data characteristics.
• Suitable assessments for evaluating different imputation approaches are discussed.

Conclusions:

• Effective imputation of missing values is crucial for reliable gene expression analysis and cancer prediction.
• Understanding algorithm classifications aids scientists in choosing optimal methods.
• This review provides a resource for adapting imputation techniques to diverse gene expression datasets.