A quantitative proteomics dataset for assessment and prediction of low dose X-ray radiation exposure in mice.

This paper presents two comprehensive quantitative proteomics datasets derived from mouse skin samples exposed to varying doses, dose rates, and time points of X-ray radiation, providing high-quality data to facilitate the discovery of biomarkers and the development of machine learning models for estimating radiation exposure.

The Gist

Imagine your body is like a highly complex city, and when it gets hit by invisible "X-ray rain," the city's workers (proteins) start reacting in specific, tell-tale ways. Usually, if you want to know how much radiation someone was exposed to, you need a physical badge or meter. But what if that badge is missing? This paper is about creating a new kind of "detective kit" that reads the city's workers to figure out exactly how much "rain" fell, even without the original meter.

Here is how the researchers built this kit:

The Experiment: A Controlled Storm
The scientists took tiny samples of skin from mice and subjected them to controlled doses of X-rays, like setting up different levels of a storm. They didn't just test one thing; they created two massive "weather logs":

1. The Quick Test (Experiment 1): They zapped 96 mice samples with doses ranging from zero to a high level, delivered either slowly or very quickly. They checked the results 6 days after the "storm" passed.
2. The Long-Term Watch (Experiment 2): This was the big one, involving 936 samples. They exposed mice to doses ranging from zero to 100, using different speeds of delivery, and then watched how the samples changed over a long period—from just one week up to five months later.

The Detective Work: Reading the Protein "Fingerprints"
Once the samples were collected, the team didn't just look at them with a microscope. They used a high-tech machine (mass spectrometry) that acts like a super-precise librarian. It sorted through millions of tiny molecular "books" (proteins) to see which ones were more or less common after the radiation.

To make sure the results were trustworthy, they included "reference samples"—think of these as the control group or the "standard ruler" used to make sure the measuring tape wasn't stretched or broken. They used a standardized, automated process to ensure every single sample was treated exactly the same way, minimizing human error.

The Result: A Massive Library for Future Use
The paper doesn't just say "we did this"; it actually opens the doors to a giant digital library. All the raw data, the cleaned-up numbers, and the lists of protein changes are available for anyone to download.

Think of this dataset as a massive instruction manual for a computer program. Scientists can now use this data to train computers (machine learning) to recognize the specific "fingerprint" left by X-rays. The goal is to help researchers understand exactly how radiation changes the biology of mice, which the authors hope will eventually help inform how we understand these changes in humans too.

In short, this paper provides the raw ingredients and the recipe for a new way to detect invisible radiation exposure by reading the molecular "echoes" left behind in skin tissue.

Technical Summary

Technical Summary

Problem Statement
The accurate estimation of ionizing radiation exposure is critical, particularly in scenarios where physical dosimeters are unavailable. While ionizing radiation is known to induce molecular responses, there is a need for comprehensive, high-quality datasets that characterize these responses across varying exposure conditions to facilitate reliable dose estimation and biomarker discovery.

Methodology
This study presents two large-scale quantitative proteomics datasets derived from mouse dorsal skin punch samples. The experimental design systematically varied three key parameters: radiation dose, dose rate, and post-exposure time.

• Experiment 1: Focused on acute exposure effects, comprising 96 samples (including 16 reference samples). These were collected 6 days after exposure to doses ranging from 0 to 75 cGy, delivered at two distinct dose rates: 30 cGy/min and 300 cGy/min.
• Experiment 2: Designed to capture long-term temporal dynamics, comprising 936 samples (including 236 reference samples). These samples were exposed to doses between 0 and 100 cGy at lower dose rates (3 cGy/min and 28 cGy/min) and harvested across a broad time window from 7 to 150 days post-exposure.

All samples underwent a standardized processing workflow involving automated bead-based digestion followed by data-independent acquisition (DIA) mass spectrometry. To ensure data integrity and quality, the workflow incorporated multiple pooled reference sample types, process controls, and system suitability standards.

Key Contributions and Results
The primary contribution of this work is the generation and public release of two extensive proteomics datasets available via ProteomeXchange. The data is provided at multiple levels of processing, ranging from raw files to normalized peptide- and protein-level abundance matrices. These resources are specifically structured to support downstream applications, including biomarker discovery and machine learning model training for exposure assessment. The inclusion of diverse controls and reference samples ensures the high quality and reproducibility of the data.

Significance
The authors posit that this dataset will facilitate the generation of new insights into the specific biological changes and molecular signatures resulting from X-ray exposure in mice. Furthermore, the work claims potential utility in informing future studies regarding human radiation exposure, serving as a foundational resource for developing predictive models when physical dosimetry is not feasible.