NEW: annotating real spectra vs calculated one #314
Conversation
There was a problem hiding this comment.
This module currently contains the conversion of a flat (annotated) library to a dense library. In the meantime we have implemented a more optimized and clean version in alphabase.
See here: create_dense_matrices
| # Loss type mappings - these map loss type numeric codes to their string representations | ||
| # Used in annotate.py and flat.py | ||
|
|
||
| LOSS_NUMBER_TO_TYPE = {0: "", 17: "_NH3", 18: "_H2O", 98: "_modloss"} |
There was a problem hiding this comment.
This is now covered in https://github.com/MannLabs/alphabase/blob/main/alphabase/peptide/fragment.py
There was a problem hiding this comment.
How much redundency is there between these two modules?
There was a problem hiding this comment.
i'll try to minimize redundancy, thank you for pointing it out
| ---------- | ||
| speclib_flat : SpecLibFlat | ||
| A spectral library flat object containing precursor information. | ||
| raw_data : MzMLReader | ThermoRawData |
There was a problem hiding this comment.
I think the most accurate thig would be to use a MsDataBase from alpharaw here as annotation type.
There was a problem hiding this comment.
can't use alpharaw, otherwise we get circular dependencies
| ) | ||
|
|
||
|
|
||
| def _get_dense_column( # noqa: PLR0913 |
| return "_".join(items) | ||
|
|
||
|
|
||
| def _add_frag_column_annotation( |
| ) | ||
|
|
||
|
|
||
| def _assign_to_dense( |
| ) | ||
| from alphabase.spectral_library.flat import SpecLibFlat | ||
|
|
||
| UNANNOTATED_TYPE = 255 |
There was a problem hiding this comment.
can you move this to alphabase/peptide/fragment.py?
| return outlib_flat | ||
|
|
||
|
|
||
| def calculate_pif(spectrum_peak_df: pd.DataFrame) -> float: |
There was a problem hiding this comment.
What do you think about moving them into a metrics.py module? similar for gini etc.
| ) | ||
|
|
||
|
|
||
| def _sequence_coverage_metric(flatlib: SpecLibFlat) -> None: |
| ) | ||
|
|
||
|
|
||
| def _sequence_gini_metric(flatlib: SpecLibFlat) -> None: |
| ) | ||
|
|
||
|
|
||
| def _normalized_count_metric(flatlib: SpecLibFlat) -> None: |
| return np.median(valid_values) | ||
|
|
||
|
|
||
| def _mass_accuracy_metric(flatlib: SpecLibFlat) -> None: |
| mass_error_ppm=mass_error_ppm, | ||
| ) | ||
|
|
||
| matched_precursor_df.loc[i, "pif"] = calculate_pif(spectrum_peak_df) |
There was a problem hiding this comment.
Sorry, that I'm asking you to fix my mistakes here :D
could you change the calulate_pif() function so it's structured and applied in the same way as the other metrics?
So it's called after annotation and operates on the precursor df.
| ) | ||
|
|
||
|
|
||
| def add_dense_lib( |
There was a problem hiding this comment.
what do you think about adding this function directly to the speclib flat object?
so we can call SpecLibFlat.add_dense_lib().
I think this fits the ab pattern better.
| @@ -261,16 +262,16 @@ def get_full_charged_types(self, frag_df: pd.DataFrame) -> list: | |||
| # Now if we have a fragment type that is a,b,c we should have the corresponding x,y,z | |||
|
|
|||
| corresponding = {"a": "x", "b": "y", "c": "z", "x": "a", "y": "b", "z": "c"} | |||
| loss_number_to_type = {0: "", 18: "_H2O", 17: "_NH3", 98: "_modloss"} | |||
There was a problem hiding this comment.
get_full_charged_types() is deprecated, so there is no need to do any changes here.
| # Loss type mappings - these map loss type numeric codes to their string representations | ||
| # Used in annotate.py and flat.py | ||
|
|
||
| LOSS_NUMBER_TO_TYPE = {0: "", 17: "_NH3", 18: "_H2O", 98: "_modloss"} |
There was a problem hiding this comment.
I think this can be removed as it's not needed by the new AB based implementation :).
@GeorgWa added functionality to compare experimental spectra with calculated ones, so all credit to him.
taken from https://github.com/MannLabs/metaptcm/blob/cbf579b52041d96636dd04c29d4855e5391bd136/metaptcm/utils/annotation.py so @mschwoer already reviewed this
Feel free to suggest a different placement within the package, also added some noqas.