LTR: Linear Cross-Platform Integration of Microarray Data.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Microarray experiments are increasing in scale and complexity.
• Datasets from different platforms or centers often contain platform- and batch-specific biases.
• Existing techniques struggle to effectively remove these biases.

Purpose of the Study:

• To develop and evaluate a novel algorithm, Linear modeling (LTR), for removing biases in microarray data.
• To specifically utilize replicate hybridizations as an experimental control for bias correction.
• To assess the independence of LTR's effectiveness from data pre-processing methods.

Main Methods:

• LTR is a linear-modelling-based algorithm designed to learn relationships between microarray batches.
• The algorithm uses replicate hybridizations of samples across different platforms or batches.
• LTR was tested on a benchmark dataset using different Affymetrix microarray platforms and six pre-processing algorithms.

Main Results:

• LTR successfully removed significant biases between two different Affymetrix microarray platforms.
• The effectiveness of LTR was consistent across six different data pre-processing algorithms.
• Sample-size experiments demonstrated that as few as three replicate hybridizations can substantially reduce bias.

Conclusions:

• LTR is an effective algorithm for removing platform- and batch-specific biases in microarray data.
• The algorithm is robust and its performance is independent of the chosen pre-processing method.
• LTR offers a valuable tool for improving the comparability of microarray datasets, with an R library available.