Skip to content

Latest commit

 

History

History
73 lines (53 loc) · 2.34 KB

File metadata and controls

73 lines (53 loc) · 2.34 KB

Quantification - Usage Guide

Overview

Convert raw mass spectrometry signals into protein abundance estimates using label-free, isobaric, or metabolic labeling strategies.

Prerequisites

pip install numpy pandas scipy
# R packages: BiocManager::install(c("MSstats", "DEP", "MSnbase"))

Quick Start

Tell your AI agent what you want to do:

  • "Normalize my MaxLFQ intensities using median centering"
  • "Process TMT reporter ion intensities from my experiment"
  • "Impute missing values using KNN for my protein matrix"

Example Prompts

Normalization

"Apply median centering normalization to my protein intensity matrix"

"Use quantile normalization to correct for batch effects between runs"

"Normalize my TMT data using the internal reference channel"

Missing Value Handling

"Impute missing values using KNN for MAR pattern and MinProb for MNAR"

"Filter proteins with more than 50% missing values, then impute the rest"

"Analyze the missing value pattern and recommend an imputation strategy"

Label-Free Quantification

"Calculate MaxLFQ intensities from peptide-level data"

"Summarize peptide intensities to protein level using top3 method"

TMT/iTRAQ Processing

"Extract TMT reporter ion intensities and correct for isotope impurity"

"Normalize across TMT plexes using the bridge channel"

What the Agent Will Do

  1. Load protein/peptide intensity matrix
  2. Log2-transform raw intensities
  3. Apply appropriate normalization method
  4. Identify missing value pattern (MCAR/MAR/MNAR)
  5. Impute missing values with suitable method
  6. Generate QC metrics (CV, correlation)

Normalization Methods

Method Description Use Case
Median centering Shift to common median General purpose
Quantile Force identical distributions Strong batch effects
LOESS Local regression Non-linear effects
VSN Variance stabilization Heteroscedastic data

Missing Value Handling

Type Method
MCAR Mean/median imputation
MAR KNN imputation
MNAR (low abundance) MinDet, MinProb, left-censored

Tips

  • Always log2-transform before normalization
  • Check CV across replicates (technical <20%, biological <40%)
  • Use PCA to verify normalization removed batch effects
  • Document imputation method - it affects downstream statistics