PTM Analysis - Usage Guide

Overview

Localize and quantify post-translational modifications (phosphorylation, acetylation, ubiquitination, glycosylation) as three stacked inferences: which sites the enrichment chemistry actually captured, where on the peptide the modification sits (with its own false localization rate), and whether an abundance change survives subtracting the protein-level change. The central decision is never trusting a phospho-only fold-change as "regulation" without protein-level adjustment.

Prerequisites

pip install pandas numpy scipy
# R packages: BiocManager::install(c("MSstatsPTM"))
# Kinase activity: KSEAapp (CRAN) or PTM-SEA / ssGSEA2.0 (GitHub broadinstitute/ssGSEA2.0)
# CLI search engines (upstream): MaxQuant, FragPipe/MSFragger, DIA-NN, Spectronaut

Quick Start

Tell your AI agent what you want to do: - "Load my MaxQuant Phospho (STY)Sites.txt, expand multiplicity, and keep class I sites" - "Adjust phosphosite changes for protein abundance using MSstatsPTM and a paired global proteome" - "Build a kinase-motif logo using an experiment-matched background, not the whole proteome" - "Infer which kinases are active with KSEA from my site fold-changes" - "Check whether my diGly sites are confounded by NEDD8/ISG15 or an iodoacetamide artifact"

Example Prompts

Site Identification and Localization

"Filter Phospho (STY)Sites.txt to class I (localization probability >= 0.75) and report the residue distribution"

"Expand the MaxQuant site-table multiplicity into Intensity1/2/___3 before any quantification"

"Estimate an empirical global false localization rate for my phosphosites"

Protein-Adjusted Quantification

"Use MSstatsPTM groupComparisonPTM and call only ADJUSTED.Model hits as regulated"

"Show me which apparent site changes are actually driven by protein abundance"

"Compare phosphosite changes after drug treatment, adjusting for protein-level stabilization"

Other PTMs

"Treat my K-GG data as ubiquitin plus NEDD8 plus ISG15 and flag the chemistry confounds"

"Map acetylation sites allowing four or more missed cleavages because acetyl-K blocks trypsin"

"Disambiguate glycosite N->D from spontaneous deamidation"

Motif and Kinase Activity

"Run motif analysis with an experiment-matched S/T/Y background and render a sequence logo"

"Run KSEA or PTM-SEA to infer active kinases and report z-scores with substrate counts"

What the Agent Will Do

Frame the question as enrichment -> localization -> quantification and identify which layer the user is asking about
Load the search-engine site output, drop Reverse/contaminant, and filter localization probability to class I
Expand MaxQuant multiplicity to a long, multiplicity-resolved site matrix before any statistics
Run MSstatsPTM with a paired global proteome and report PTM, PROTEIN, and ADJUSTED models, calling only ADJUSTED hits regulated
Perform motif analysis against an experiment-matched background and kinase-activity inference with a curated prior
Report three numbers (peptide FDR, localization probability, global FLR) and triage hits by functional evidence

Tips

Always acquire a paired global (unenriched) proteome; without it, "regulated site" claims are unfalsifiable.
A between-method or between-lab difference is a chemistry hypothesis first (TiO2 vs Fe-IMAC mono/multi bias), biology second.
Quantify on multiplicity columns (three underscores), never the collapsed base Intensity.
For ubiquitinomes, confirm chloroacetamide alkylation and remember K-GG is not ubiquitin-specific.
Build motif backgrounds from the matched dataset; whole-proteome backgrounds just rediscover disordered-region composition bias.
Kinase-activity inference is prior-limited; simple z-score matches sophisticated methods, so invest in the substrate prior.
Most sites have no annotated function; triage by conservation, stoichiometry, and Ochoa functional score before claiming biology.

peptide-identification - Identify modified peptides and run open/variable-mod search
quantification - Underlying protein-level quant feeding the MSstatsPTM PROTEIN dataset
differential-abundance - Moderated testing on the protein-level intensity matrix
pathway-analysis/gsea - Enrichment scoring of regulated-site protein lists and PTM-SEA-style signatures
data-visualization/sequence-logos - Render motif logos from the foreground/background windows