elucidatainc / corna Goto Github PK

Try running this file
auto_detect_indistinguish_dual_label copy.txt

For Glutathione the expected output is:
40% unlabeled and 60% 13C-3 15N-1 which I get if I give the indistinguishable isotopes manually i.e. {'C': ['H','O'], 'N':['S']}

But not with auto-detect. The reason being this function:

def make_correction_matrix(trac_atom, formuladict, na_dict, indist_elems):
"""create matrix M such that Mx=y where y is the observed isotopic distribution
and x is the expected distribution of input labels

atom_bag: dict of element:number of atoms in molecule (e.g. {'C':2,'O':1,'H':6})
label_elem: element with input labeling
indist_elems: elements with identical mass shift
na_dict: dict of element:expected isotopic distribution
:TODO This function relates to issue NCT-247. Need to change the function
in more appropriate way.
"""
M = make_expected_na_matrix(formuladict.get(trac_atom, 0), na_dict[trac_atom])
for e in indist_elems:
    if e in formuladict:
        e1 = ISOTOPE_NA_MASS[KEY_ELE][e]
        M = add_indistinguishable_element(M, formuladict[e1], na_dict[e])
return pinv(M)

So, the if statement would reject indistinguishable isotopes like 'O17','O18' and 'S33' because these won't be in the formula. And I tested it is giving the same result as what you would get if manual input was 'C':['H'] because 'H' remains the same in both cases.

1. Ipython notebook
2. Use a publicly available dataset for correction

Currently, one workflow is using olmonk and other some inbuilt validation. If olmonk is to be used across everything, it should be made optional such that if the user doesn't have olmonk, it doesn't stop the workflow only stops the validation.

This demo should have the following:

1. Be an Ipython notebook
2. Use publicly available raw files like used for Polly IsoCorrect demo.
3. Include demo for - one label, two label, indistinguishible isotopes, auto-indistinguishible isotopes

Such dependencies should be optional