G-stack modulated probe intensities on expression arrays - sequence corrections and signal calibration

Area of Science:

• Genomics
• Bioinformatics
• Molecular Biology

Background:

• Expression microarrays measure mRNA abundance using probe intensities.
• Probes with consecutive guanine (G) runs exhibit artificially high intensities, misrepresenting target concentrations.
• This G-bias necessitates correction for reliable downstream expression analysis.

Purpose of the Study:

• To investigate the impact of guanine (G) runs on probe intensities in GeneChip expression arrays.
• To develop and propose a novel model for correcting probe intensity bias caused by poly-G motifs.

Main Methods:

• Applied positional-dependent sensitivity models to analyze probe intensities against sequence motifs.
• Evaluated nearest-neighbor (NN) and next-nearest-neighbor (NN+GGG) models for sequence motif effects.
• Assessed the efficacy of existing preprocessing methods (vsn, RMA, dChip, MAS5, gcRMA) in removing G-bias.

Main Results:

• Runs of three or more consecutive guanines, especially at the 3' end ((GGG)1-effect), significantly increase probe intensities.
• This bias affects both perfect match and mismatch probes and is more pronounced in later GeneChip generations and with T7-primer amplification.
• Standard preprocessing methods inadequately correct for G-bias, highlighting the need for specialized algorithms.

Conclusions:

• Positional and motif-dependent models accurately capture sequence effects on oligonucleotide probe intensities.
• A novel NN+GGG hybrid model is proposed to correct intensity bias from poly-G motifs.
• This model can be implemented as a single-chip calibration algorithm for pre-correction before standard microarray data preprocessing.