A0412
Title: Outlier detection in mass-spectrometry data using the conformal prediction framework
Authors: Soohyun Ahn - Ajou University (Korea, South) [presenting]
Abstract: Quality control (QC) in mass spectrometry (MS) data is essential for reliable biomarker discovery and the study of complex biological systems. Traditional methods often focus only on either sample or peak outlier detection and may use subjective criteria or uniform thresholds based on asymptotic distributions, which do not adequately capture the data's unique characteristics. This study introduces a novel approach utilizing conformal prediction for outlier detection in MS data analysis. This new method identifies outlier samples and peaks simultaneously using data-driven and distribution-free principles. Extensive numerical evaluations and comparisons with existing methods reveal its superior diagnostic performance.
