Change in Isotope Pattern Generator

[dlon450]

Describe the problem you encountered Hi OpenMS team, just wanted to ask about the difference between pyopenms versions 3.0.0 and 2.7.0 regarding the CoarseIsotopePatternGenerator and FineIsotopePatternGenerator functions as the MZ and intensity values seem noticeably different.

To Reproduce Here's one example. In version 3.0.0:

>>> find_isotope_pattern_coarse('C636H1007N167O178S5PtH-2', peak=14194.32)         
(array([14189.31094916, 14190.314304  , 14191.31765884, 14192.32101367,
       14193.32436851, 14194.32772335, 14195.33107819, 14196.33443303,
       14197.33778786, 14198.3411427 ]), array([0.00363849, 0.01054412, 0.02373276, 0.04367016, 0.0680778 ,
       0.09226516, 0.11084374, 0.11982168, 0.11793806, 0.10671093]))

Whereas in version 2.7.0:

>>> find_isotope_pattern_coarse('C636H1007N167O178S5PtH-2', peak=14194.32)         
(array([14189.31726057, 14190.32061541, 14191.32397025, 14192.32732509,
       14193.33067993, 14194.33403476, 14195.3373896 , 14196.34074444,
       14197.34409928, 14198.34745411]), array([0.04367016, 0.0680778 , 0.09226516, 0.11084373, 0.11982169,
       0.11793808, 0.10671095, 0.08945209, 0.06992197, 0.05124441]))

Function (replace CoarseIsotopePatternGenerator with FineIsotopePatternGenerator to reproduce with hyperfine isotope distribution):

from pyopenms import EmpiricalFormula, CoarseIsotopePatternGenerator

def find_isotope_pattern_coarse(formula_str: str, peak=None, interval=5.):
    '''
    Return coarse theoretical distribution of intensities
    '''
    seq_formula = EmpiricalFormula(formula_str)
    isotopes = seq_formula.getIsotopeDistribution( CoarseIsotopePatternGenerator() )
    isotopes_container = isotopes.getContainer()

    if peak is not None:
        lower, upper = binarySearchInterval(isotopes_container, np.floor(peak - interval), np.ceil(peak + interval))
        isotopes_container = isotopes_container[lower:upper]

    o = np.transpose([[iso.getMZ(), iso.getIntensity()] for iso in isotopes_container])
    return o[0], o[1]

def binarySearchInterval(isotopes, lbound, ubound):
    '''
    Find isotopes with masses in given interval
    '''
    low = 0
    high = len(isotopes) - 1

    lower_idx = binarySearch(isotopes, low, high, lbound)
    upper_idx = binarySearch(isotopes, lower_idx, high, ubound) - 1
    return lower_idx, upper_idx

def binarySearch(isotopes, low, high, X):
    '''
    Search for index position of mass X in isotopes
    '''
    mid = 0
    while low <= high:
        mid = (high + low) // 2
        mass = isotopes[mid].getMZ()
        if mass < X:
            low = mid + 1
        elif mass > X:
            high = mid - 1
        else:
            return mid
    return high + 1

In particular, the isotope with peak intensity had an MZ of 14193.33067993 in version 2.7.0 (the MZ desired for our work) whereas in version 3.0.0 has an MZ of 14196.33443303. Does anyone know what might be causing this? Thanks!

System information:

• OS: Windows
• OS Version: 11
• Python version: 3.9.7
• How did you install pyopenms? Please cross a box with an X.
  • [ ] conda::bioconda: Specify your conda/mamba command and/or your environment with conda list.
  • [ ] conda::openms: Specify your conda/mamba command and/or your environment with conda list.
  • [x] pip
  • [ ] nightly wheel
  • [ ] built from source

[jpfeuffer]

Hi! Thanks for reporting!

I suspect it is indeed a bug introduced during the correction of our definition of monoisotopic (most abundant instead of lightest).

Can you confirm by checking a formula without elements whose monoisotope is not the lightest (in your case probably without platinum)? If my theory is correct those patterns should not be affected.

[jpfeuffer]

By the way, FineIsotope should not be affected (which is also what I am seeing when I run your code).

[dlon450]

Thanks for your response. Without platinum the function returns the same value (your theory is correct)! You are also right that the hyperfine generator is not affected.

[dlon450]

Hey @jpfeuffer, thank you very much for fixing this issue. Just wondering how I would install the new update? Is it fine just to pip install pyopenms? Or will I need to wait for the next version?

[jpfeuffer]

Hi! For now you will need to use our nightly pypi package to try the fix: https://pyopenms.readthedocs.io/en/latest/user_guide/installation.html#nightly-ci-wheels

pip install --index-url https://pypi.cs.uni-tuebingen.de/simple/ pyopenms