Related Experiment Video
Updated: Jul 9, 2026
Deep Proteome Profiling by Isobaric Labeling, Extensive Liquid Chromatography, Mass Spectrometry, and Software-assisted Quantification
Published on: November 15, 2017
Advancing proteomic discovery through optimized multi-stage scoring and deep learning-enhanced open search
Chen Qian1,2, Kaifei Wang1,2, Pengzhi Mao1,2
1Institute of Computing Technology, Laboratory of Intelligent Information Processing of Chinese Academy of Sciences (CAS), Beijing, 100190, China.
Motivation:
Protein search engines are essential for interpreting mass spectrometry data into biological insight. Current tools often face limitations in sensitivity when analyzing complex modern datasets, and lack a unified framework that effectively integrates deep learning features for both restricted and open searches, especially for scenarios aimed at discovering unknown modifications.
Results:
We present pFind+, a high-performance search engine for data-dependent acquisition (DDA) proteomics, extending pFind. It introduces an enhanced raw scoring that delivers substantially improved pre-filtering ability, while recovering most of the computational overhead through a tailored acceleration strategy. Coupled with an enhanced rescoring framework that effectively integrates deep learning features, pFind+ uniquely supports high-sensitivity, DL-enhanced open search, enabling comprehensive PTM discovery while incorporating hardware-aware inference optimizations for practical deployment. Evaluations across diverse datasets demonstrate its superior sensitivity, with gains of 12.7%-29.3% (average 17.9%) in restricted search and 8.0%-38.4% (average 25.8%) in open search over the best existing tools.
