Machine learning and related approaches in transcriptomics.

Area of Science:

• Bioinformatics
• Genomics
• Computational Biology

Background:

• Transcriptomic data generation is no longer the primary bottleneck in analytical pipelines.
• Advancements in transcriptome profiling have shifted focus towards sophisticated data analysis methods.
• Machine learning (ML) offers efficient solutions for analyzing large and complex human transcriptome datasets.

Purpose of the Study:

• To bridge the knowledge gap for bioinformaticians regarding ML in human transcriptomics.
• To provide a comprehensive overview of ML types and techniques applicable to transcriptome analysis.
• To highlight recent (last five years) applications of ML in human transcriptome investigations.

Main Methods:

• Review of recent scientific literature (primarily last five years) on ML in human transcriptomics.
• Categorization of general and specific ML techniques.
• Analysis of computational aspects including data pre-processing, task formulation, model performance, and validation.

Main Results:

• Demonstration of ML techniques integrated into human transcriptome analysis pipelines.
• Comparison of ML-based outcomes with traditional non-ML analytical tools.
• Identification of key computational considerations for implementing ML in transcriptomics.

Conclusions:

• ML presents a powerful approach to enhance the analysis of human transcriptome data.
• Understanding ML methodologies is crucial for bioinformaticians to leverage these advanced tools.
• This review serves as a practical guide for applying ML in human transcriptomic research.