化学
等压标记
等压法
片段(逻辑)
聚糖
质谱法
工作流程
色谱法
串联质谱法
天际线
质量
计算生物学
蛋白质组学
数据挖掘
Python(编程语言)
人工智能
复制
鸟枪蛋白质组学
计算机科学
模式识别(心理学)
鉴定(生物学)
串联质量标签
无标记量化
分辨率(逻辑)
特征选择
数据缩减
判别式
管道(软件)
降维
吉祥物
定量蛋白质组学
限制
特征(语言学)
生物系统
反褶积
自动化
作者
Vishal Sandilya,Sherifdeen Onigbinde,Sarah Sahioun,Abiodun Adewolu,Ronny Barrera,Yehia S. Mechref
标识
DOI:10.1021/acs.analchem.6c04973
摘要
Abstract Targeted parallel reaction monitoring (PRM) mass spectrometry enables sensitive and reproducible glycan quantification, but data analysis for permethylated N-glycans currently relies on manual adaptation of proteomics tools such as Skyline combined with GlycoWorkbench for transition curation. This workflow requires extensive user intervention for fragment validation, isobaric disambiguation, and peak integration, limiting throughput and introducing operator-dependent variability. We present GlycanPRMQuant, an open-source Python pipeline that integrates precursor-glycan matching, fragment-level MS/MS annotation, automated isobaric conflict resolution via diagnostic fragment fingerprinting, and fragment-confirmed AUC quantification into a single reproducible workflow. The isobaric conflict-resolution step represents the primary algorithmic contribution. Competing assignments within a chromatographic feature are evaluated using a discriminative score constructed from specificity-weighted fragment intensity, effective distinct fragment support, fragment mass accuracy, chromatographic coherence, and calibrated precursor mass likelihood. Assignments are further validated by target-decoy competition with charge-aware decoy fragments and are reported only when they satisfy the fragment-evidence requirements, defeat their corresponding decoy, and meet the specified q-value threshold. Features lacking discriminating evidence are reported as ambiguous rather than assigned. Unlike Skyline-based workflows, which require per-glycan transition curation in GlycoWorkbench and manual peak boundary review, GlycanPRMQuant produces identification and quantitation outputs directly from mzML input with no user intervention, reducing processing time and eliminating a principal source of inter-analyst variability. Quantitative agreement with a manually curated Skyline reference yielded Pearson’s R of 0.992−1.000, with median CVs of 5.08% and 10.72% for bovine fetuin and ribonuclease B glycans across replicate injections.
科研通智能强力驱动
Strongly Powered by AbleSci AI