GenoToxMass Standard Operating Procedure (SOP)

Overview

This document describes the standard workflow for predicting chemical genotoxicity using the GenoToxMass platform, covering sample preparation, LC-MS/MS data acquisition, feature extraction, model invocation, result interpretation, and quality control.

It is intended for:

• Regulatory agencies
• Contract research organizations (CROs)
• Analytical laboratories

The procedure is designed to support standardized implementation, independent validation, and routine operational use.

1. Scope and Purpose

This SOP defines the complete workflow for predicting chemical genotoxicity using the GenoToxMass platform, from sample preparation to result interpretation.

2. Instrumentation and Materials

2.1 Chromatographic System

Item	Specification
Recommended column	C18 column (2.1-3.5 um particle size, 2.1 x 50 mm)
Column temperature	30 +/- 5 C
Mobile phase A	Water with 0.1% formic acid (LC-MS grade)
Mobile phase B	Acetonitrile with 0.1% formic acid (LC-MS grade)
Flow rate	0.3-1.0 mL/min
Injection volume	1-5 uL

Gradient Program

Time (min)	%B
0.0	5
1.0	5
8.0	95
10.0	95
10.1	5
12.0	5

2.2 Mass Spectrometry System

Item	Specification
Ionization source	Electrospray Ionization (ESI)
Scanning method	Select the mode exhibiting higher signal strength
Capillary voltage	5.0 kV
Ion source temperature	325 C
Desolvation gas flow	12 L/min (N2)
Cone gas flow	12 L/min (N2)
Collision energy	Ensure the molecular ion peak is retained at moderate intensity (10%-80%)
Mass range	m/z 50-1000
Resolution	>= 20,000 (FWHM) at m/z 200

2.3 Quality Control Standards

Item	Specification
System suitability test mixture	Colchicine (CAS 64-86-8), Reserpine (CAS 50-55-5), each at 1 ug/mL
Acceptance criteria	Retention time variation <= 0.2 min; mass accuracy <= 5 ppm; intensity variation <= 20% across three consecutive injections

3. Sample Preparation

3.1 Environmental Water Samples

1. Filter sample through a 0.22 um nylon membrane filter.
2. Add internal standard to a final concentration of 100 ng/mL.
3. Concentrate 100 mL sample using solid-phase extraction (SPE).
4. Condition SPE cartridge with 5 mL methanol, then 5 mL water.
5. Load sample at 5 mL/min.
6. Wash with 5 mL 5% methanol in water.
7. Elute with 5 mL methanol.
8. Evaporate to dryness under N2 at 40 C.
9. Reconstitute in 200 uL methanol:water (1:1, v/v).

3.2 Pharmaceutical Impurities

1. Dissolve drug substance in an appropriate solvent at 10 mg/mL.
2. Dilute to 10 ug/mL with methanol:water (1:1, v/v).
3. Filter through a 0.22 um PTFE syringe filter.

3.3 Food Matrices

1. Homogenize sample (5 g) with 10 mL acetonitrile.
2. Centrifuge at 10,000 x g for 10 min at 4 C.
3. Collect supernatant and perform dispersive SPE cleanup.
4. Evaporate 5 mL supernatant to dryness.
5. Reconstitute in 500 uL methanol:water (1:1, v/v).

4. Data Acquisition

4.1 Acquisition Mode

• Use data-dependent acquisition (DDA) or data-independent acquisition (DIA).
• For completely unknown samples, acquire both positive and negative ionization modes, then use the spectra from the mode with higher overall intensity for subsequent analysis.
• Acquire MS/MS spectra.
• Set dynamic exclusion to 5 s after two occurrences.

4.2 File Format Requirements

• Export raw data as .raw.
• Alternatively, use the GenoToxMass template (.xlsx) with the following required columns:

Column	Description	Example
mz_list	Comma-separated list of m/z values	121.05, 149.02, 177.05
intensity_list	Comma-separated list of corresponding intensities	100, 45, 12

5. Data Preprocessing and Feature Extraction

Software performs this section automatically.

5.1 Spectral Filtering

1. Identify the maximum intensity (MaxM) in the raw spectrum.
2. Set filtering threshold T = MaxM x P, where P = 0.01 (1%).
3. Retain only peaks with intensity >= T.

5.2 Feature Calculation (14 Parameters)

Feature	Abbreviation	Calculation / Definition
Peak Number	PN	Count of peaks after filtering
Base Peak	BP	m/z value of the most intense peak in filtered spectrum
Base Peak Proximity	BPP	m/z difference between the BP and its nearest peak
Maximum Mass	MaxM	Maximum m/z value in filtered spectrum
Maximum Mass Proximity	MaxMP	m/z difference between the MaxM and its nearest peak
Minimum Mass	MinM	Minimum m/z value in filtered spectrum
Mass Mean	MM	Mean of all m/z values in filtered spectrum
Mass Standard Deviation	MSD	Standard deviation of all m/z values
Intensity Mean	IM	Mean of all intensity values
Intensity Standard Deviation	ISD	Standard deviation of all intensity values
Intensity Density	ID	BPI / PN
Retention Time	RT	As acquired (minutes)
Collision Energy	CE	As acquired (eV)
Precursor Type	PT	0 for [M-H]-, 1 for [M+H]+, [M+Na]+, [M+K]+

Optional Generic Input Profile

If the user does not provide the three test-related parameters (RT, CE, and PT), a generic input profile may be used:

• RT = 8 min
• CE = 30 eV
• PT = positive ion mode

Important notes:

• If a specific ionization mode is selected, spectra must be acquired under that same mode.
• This fallback is not recommended unless absolutely necessary.
• Predictive performance under this mode is expected to degrade by >= 10% relative to external validation results.

6. Model Invocation

6.1 Online Platform Access

1. Navigate to https://mspredict.com/genotoxicity.
2. Download the input template from the website.
3. Prepare data following the template format.
4. Upload the completed .xlsx file.
5. Click Predict and wait for processing (typically less than 5 seconds).
6. Review the output:

• Prediction: Toxic or Non-Toxic
• Probability: value between 0 and 1

7. Result Interpretation and Decision Framework

7.1 Default Classification Threshold

• Optimal threshold tau* is determined by user need.
• Classification rule:
  • If probability >= tau* -> classify as Toxic
  • If probability < tau* -> classify as Non-Toxic

7.2 Threshold Adjustment for Specific Applications

Application Priority	Recommended Threshold	Expected Performance Trade-off
Maximize Accuracy	0.42	Balanced sensitivity/specificity
Minimize False Negatives (Safety-First)	0.30	Higher recall, lower precision
Minimize False Positives (Cost-Sensitive)	0.65	Higher specificity, lower recall
Regulatory Confirmation	0.55	Optimized for both

7.3 Confidence Scoring

Confidence Level	Probability Range
High	>= 0.8 or <= 0.2
Medium	0.6 <= p < 0.8 or 0.2 < p <= 0.4
Low	0.4 < p < 0.6

8. Quality Control and Validation

8.1 System Suitability Test

1. Inject system suitability test mixture (colchicine, reserpine).
2. Verify retention times are within +/- 0.2 min of established values.
3. Verify mass accuracy <= 5 ppm for all three compounds.
4. Run GenoToxMass prediction on all three.
5. If any prediction deviates, investigate instrument performance.

8.2 Batch Quality Control

• Include one positive control (for example, colchicine) and one negative control (for example, reserpine) in each analytical batch.
• Acceptance criteria: correct classification for both controls.
• If controls fail, reject the entire batch and re-analyze.

8.3 Inter-Laboratory Transfer

For laboratories implementing this method for the first time:

1. Analyze the reference standard set (n = 20 compounds).
2. Compare predictions with reference values.
3. Acceptable performance: >= 85% concordance.
4. Document all results for regulatory audit.

9. Documentation and Reporting

All analyses should be documented with:

1. Instrument calibration records
2. System suitability test results
3. Sample preparation logs
4. Raw data files
5. Feature extraction worksheets
6. Model prediction outputs
7. Final interpretation and decision rationale

10. Troubleshooting Guide

Problem	Possible Cause	Solution
Low peak count after filtering	High noise level	Increase injection volume; improve sample cleanup
Unexpected predictions	Poor spectral quality	Verify collision energy; check precursor ion selection
Platform unavailable	Server maintenance	Contact support

Source

Converted from the uploaded SOP PDF for GenoToxMass.