Mining the Dynamic Genome: A Method for Identifying Multiple Disease Signatures Using Quantitative RNA Expression...

Area of Science:

• Biotechnology
• Genomics
• Molecular Diagnostics

Background:

• Blood mRNA transcriptomics offers diagnostic advantages over tissue samples.
• Challenges remain in accurately measuring gene expression in whole blood due to inherent variability.
• Improved methods are needed to detect biological signatures in blood samples.

Purpose of the Study:

• To develop and validate a method for compensating collection tube bias in blood transcriptomic analysis.
• To identify a liver cancer-specific gene signature using a robust prediction model.
• To assess the performance of the developed method on independent blood sample sets.

Main Methods:

• Collection tube bias compensation was applied to identify a liver cancer gene signature.
• A prediction model using differential gene pairs was constructed to minimize confounding factors.
• The method's performance was evaluated on 157 blood samples collected in PAXgene tubes and compared to an alternative model.

Main Results:

• The developed model effectively discriminated liver cancer in both EDTA and PAXgene collected samples.
• The model compensated for collection tube bias, unlike a Weka-derived model.
• Cross-validation confirmed the procedure's accuracy in classifying samples into disease and control groups.

Conclusions:

• The developed method offers a versatile solution for blood transcriptomic investigations.
• This approach overcomes limitations in current blood-based gene testing research.
• The findings advance the potential of blood mRNA transcriptomics for diagnostics.