Wavelet-based detection of transcriptional activity on a novel Staphylococcus aureus tiling microarray

Area of Science:

• Genomics
• Bioinformatics
• Molecular Biology

Background:

• High-density oligonucleotide microarrays are crucial for genomic analysis, including transcriptional mapping and ChIP-on-chip studies.
• Tiling microarray data analysis aids in discovering novel transcripts and assessing differential gene expression.
• Despite advancements like next-generation sequencing, microarrays remain valuable for small genomes and custom genomic region analysis due to cost-effectiveness and accurate expression quantification.

Purpose of the Study:

• To introduce a novel wavelet-based method, ZCL (zero-crossing lines), for combined denoising and segmentation of tiling microarray signals.
• To evaluate the ZCL method's performance against established algorithms using public datasets.
• To demonstrate the utility of ZCL in analyzing specific biological questions, such as identifying gene expression changes in microbial mutants.

Main Methods:

• The ZCL method utilizes SUREshrink for denoising and Continuous Wavelet Transform (CWT) for detecting transcriptionally active regions.
• Transition detection in CWT is achieved through thresholding zero-crossing lines.
• The algorithm was applied to Saccharomyces cerevisiae datasets and compared with pseudo-median sliding window (PMSW) and structural change model (SCM) algorithms.

Main Results:

• The ZCL method demonstrated superior performance in positive predictive value (PPV) while maintaining comparable sensitivity to PMSW and SCM.
• ZCL offers reduced computation time compared to model-based methods and is comparable to filter-based approaches.
• Application to a Staphylococcus aureus mutant dataset successfully identified transcripts with decreased expression in the absence of the sigma B transcription factor.

Conclusions:

• The ZCL method is highly suitable for analyzing tiling signals, effectively revealing transcriptional activity obscured by noise.
• The algorithm features automatic parameter selection and is amenable to parallel implementation, enhancing its practical applicability.
• The ZCL method proved valuable for quantifying and analyzing differential gene expression in the S. aureus transcriptome, showcasing its utility in biological research.