xushuwenn/redact/matchms

from matchms.importing import load_from_msp
# Load MSP file
spectra = list(load_from_msp("/root/input.msp"))

for spectrum in spectra:
    # IMPORTANT: Use 'compound_name' not 'name' for MSP 'Name:' field
    name = spectrum.get("compound_name", "Unknown")
    peaks_mz = spectrum.peaks.mz          # Array of m/z values
    peaks_intensities = spectrum.peaks.intensities  # Array of intensities
    num_peaks = len(spectrum.peaks.mz)    # Number of peaks

# Total number of spectra
total_spectra = len(spectra)
# Find spectrum with most peaks
max_peaks = 0
spectrum_with_most = None
for spectrum in spectra:
    peak_count = len(spectrum.peaks.mz)
    if peak_count > max_peaks:
        max_peaks = peak_count
        # Use 'compound_name' for the MSP Name: field
        spectrum_with_most = spectrum.get("compound_name", "Unknown")
# Total peaks across all spectra
total_peaks = sum(len(spectrum.peaks.mz) for spectrum in spectra)