Empirical methods for controlling false positives and estimating confidence in ChIP-Seq peaks.

Area of Science:

• Genomics
• Molecular Biology
• Bioinformatics

Background:

• High-throughput sequencing, including ChIP-Seq, identifies genomic elements like transcription factor binding sites and epigenetic modifications.
• Understanding these elements is crucial for reconstructing gene regulatory networks involved in development and disease.

Purpose of the Study:

• To develop and present algorithms and software for analyzing ChIP-Seq data.
• To reduce false positives and estimate confidence in ChIP-Seq peak identification using control input data.

Main Methods:

• Comparison of various algorithms using simulated spike-in datasets.
• Utilizing control input data and a normalized difference score to improve peak recovery.
• Employing binomial p-value/q-value and empirical FDR for reliable confidence estimation.

Main Results:

• Control input data and normalized difference score more than doubled ChIP-Seq peak recovery at a 5% false discovery rate (FDR).
• Binomial p-value/q-value and empirical FDR proved more reliable for predicting true FDR than global Poisson p-value.
• Successfully reanalyzed NRSF ChIP-Seq data without prior knowledge of binding sites or motifs.

Conclusions:

• The developed methods effectively reduce false positives and enhance confidence in ChIP-Seq data analysis.
• These advancements are available as part of a larger, open-source software package.
• The approach allows for reliable ChIP-Seq analysis without requiring prior knowledge of the ChIP target.