WT-2 0.0.1

WT2 Mass spectrometry data processing library

1. Installation

WT_2 requires Python 3.10. We recommend creating a virtual environment using conda:

conda create --name WT2 python=3.10
conda activate WT2
pip install WT_2

2. Usage Guide

2.1 Data Preparation

• Each sample requires 10 raw MGF files and MS-DIAL processing results.
• Place all MGF files and MS-DIAL results for the same sample into a folder named after the sample.
• Multiple samples should be organized in separate folders.
• Refer to the sample1 folder in test_data as an example.

2.2 Peak Process

from WT_2 import MultiprocessingManager

sample_folder = "./test_data/sample1" 
out_dir = "./test_data/sample1" 

manager = MultiprocessingManager(
    outer_max_workers=1,
    inner_max_workers=8,
    mgf_folder=sample_folder,
    out_dir=out_dir,
)
manager.process_mgf_files()

• Parameters:
  • process_mgf_files() will automatically create a result folder in out_dir to store peak extraction and clustering results.
  • sample_folder: Sample directory path.
  • out_dir: Output directory path.
  • outer_max_workers: Number of outer processes for MGF file processing (default: 1).
  • inner_max_workers: Number of inner processes for m/z processing (default: 8).
  • RT_start: Left boundary of retention time (RT) range (seconds).
  • RT_end: Right boundary of RT range (seconds).
  • fp_wid: Peak detection window width (default: 6).
  • fp_sigma: Peak detection sigma (default: 2).
  • fp_min_noise: Noise threshold for peak detection (default: 200).
  • group_wid: Peak clustering window width (default: 6).
  • group_sigma: Peak clustering sigma (default: 0.5).

2.3 Peak Deduplication

from WT_2 import Deduplicator


sample_name = os.path.basename(sample_folder)
msdial_path = "./test_data/sample1/CRL_SIF_1_Q1_peak_df.csv"


deduplicator = Deduplicator(
    peak_result_dir=os.path.join(out_dir, "result"),
    msdial_out_path=msdial_path,
    sample_name=sample_name,
    useHrMs1=False,
    HrMs1model_path=None
)
deduplicator.remove_msdial_duplicate()
peak_outpath, group_outpath = deduplicator.filter_p3_group()

• Parameters:
  • peak_result_dir: Directory containing peak extraction results (default: result folder in out_dir).
  • msdial_out_path: Path to MS-DIAL input file.
  • sample_name: Sample name (default: folder name).
  • useHrMs1: Whether to use high-resolution MS1 data (True for high-res, False for low-res, default: False).
  • HrMs1model_path: Path to high-resolution MS1 prediction model. If None, the pretrained model will be downloaded automatically. Pretrained model available at test_data/models/HrMs1.pth or Hugging Face

2.4 Metabolite Identification

from WT_2 import MspGenerator, MspFileLibraryMatcher
import pandas as pd

# Generate MSP file from deduplicated P3 group results
df = pd.read_csv(group_outpath)
out_msp_path = os.path.join(os.path.dirname(group_outpath), sample_name + ".msp")
msp_generator = MspGenerator(df, out_msp_path, useHrMs1=True)
msp_generator.generate()

# Library matching (requires MSP-format library)
out_match_path = os.path.join(os.path.dirname(group_outpath), sample_name + "_match_library_out.csv")

library_matcher = MspFileLibraryMatcher(
    query_msp_path=out_msp_path,
    library_msp_path="./test_data/library_msp",
    out_path=out_match_path
)
library_matcher.match()

• Parameters:
  • query_msp_path: MSP file generated from P3 group results (standard MSP format).
  • library_msp_path: Path to reference MSP library.
  • out_path: Output path for matching results.

2.5 Metabolite quantification

from WT_2 import SampleQuantity


quantity_folder = "./test_data/quantity_prepared"


quantifier = SampleQuantity(
    quantity_folder=quantity_folder,
    quantity_out_path=quantity_folder,
    ref_file=None,
    useHrMs1=True,
    uesSampleAligmentmodel=True,
    SampleAligmentmodel_path=None
)
quantifier.quantity_processor()

• Parameters:
  • quantity_folder: Directory containing deduplicated P3 results for all samples.
  • quantity_out_path: Output directory.
  • ref_file: Reference sample file (uses first sample as reference if None).
  • useHrMs1: Whether to use high-resolution MS1 data (requires prior high-res processing).
  • uesSampleAligmentmodel: Enable sample alignment model.
  • SampleAligmentmodel_path: Path to sample alignment model. If None, downloads pretrained model. Pretrained model available at test_data/models/samplealigment.pth or Hugging Face