Reading and subsampling MGF file

[ekopylova]

Hello,

I would like to read a .mgf file having thousands of spectra, subsample it down to a list of specific spectra and then write the new subsampled spectra to a separate .mgf file. Is it possible to use mgf.write() in a loop where I append the spectra which match my filtering? I'm looking for something like:

    with mgf.MGF(input_mgf_fp) as reader:
        for spectrum in reader:
            params = spectrum.get('params')
            if params['feature_id'] in feature_ids:
                mgf.write(spectrum, output = output_fp);

Thanks!

[mobiusklein]

You could do almost exactly what you have there, just wrap each individual spectrum in a list or tuple:

with mgf.MGF(input_mgf_fp) as reader:
    for spectrum in reader:
        params = spectrum.get('params')
        if params['feature_id'] in feature_ids:
            mgf.write([spectrum], output_fp)

The first parameter of mgf.write is an iterable of spectra, so in order to match its expectations, we need to wrap single spectrum dict objects in something that satisfies the expectation that next(iter(obj)) -> dict. This happens to work because mgf.write doesn't close the output stream and doesn't automatically append a file header unless you tell it to.

Another option is to use a generator to abstract away the filtering logic:

def producer(reader, feature_ids):
    for spectrum in reader:
        params = spectrum.get('params')
        if params['feature_id'] in feature_ids:
            yield spectrum

 with mgf.MGF(input_mgf_fp) as reader:
    spectrum_iterator = producer(reader, feature_ids)
    mgf.write(spectrum_iterator, output_fp)

This approach is usually more generalizable, though it does make it a little bit harder to follow because the filtering logic is no longer inline at the place where it's used.

[ekopylova]

Thank you, the minor change worked perfectly!