Adduct-Based MS/MS Analysis and Visualization Pipeline

This repository contains a complete workflow for analyzing and visualizing adduct-specific MS/MS spectral similarity using R scripts.

---

Workflow Overview

1. Step1_Adduct_Sorting.R
   Organizes MS/MS data by adduct types from input spectral libraries.

2. Step2_Finding_the_Adduct_Pairs.R
   Identifies all valid adduct pairings based on identical chemical structures, the same or similar MS instrument type, and closely matched CEs (within 3 eV).

3. Step3_Similarity_Calculation.R
   Calculates modified cosine similarity between MS/MS spectra across adduct pairs.

Once Steps 1–3 are complete, you can generate specific figures using the scripts below.

---

📊 Figure Generation Scripts

• Plot_for_fig1.R: Example MS/MS spectral comparison shows two key factors: adducts and CE
• Plot_for_fig2.R: Spectral similarity analysis: Density plot of modified cosine similarity of MS/MS comparison
• Plot_for_fig3.R: Matched fragment ratios and fragment matched types analysis
• Plot_for_fig5A.R: The trend between modified cosine similarity and CE
• Plot_for_fig5B.R: The trend between modified cosine similarity and ΔCE
• Plot_for_fig5C.R: The trend between mean geometric m/z center shift values and CE levels
• Plot_for_fig6A.R: Direct spectral searching annotation results impaired by mismatched adduct forms
• Plot_for_fig6B.R: Molecular networking based annotation results impaired by mismatched adduct forms
• Plot_for_fig6C_Step1.R: Effects of adduct forms and CE thresholds on detecting PO4-related structural indicators in MS/MS
• Plot_for_fig6C_Step2.R: TPR and FPR of ML-based annotation model performance comparisons with/without excluding alkali adduct and applying CE thresholds
• Plot_for_fig6C_Step3.R: ML annotation AUC curve comparisons with/without excluding alkali adduct and applying CE thresholds

---