def peak_fits(self):
df = self._load_df_prop_from_fields("peak_fit_df")
check.df_t(df, self.peak_fit_df_schema)
# The peaks have a local frame_peak_i but they
# don't have their pan-field peak_i set yet.
df = df.reset_index(drop=True)
df.peak_i = df.index
return df
def it_radmats():
rad_df = res.radmats()
check.df_t(rad_df, SigprocV1Result.radmat_df_schema)
assert len(rad_df) == 4 * 2 * 3
# Sanity check a few
assert (rad_df[(rad_df.peak_i == 1) & (rad_df.channel_i == 1) &
(rad_df.cycle_i == 1)].signal.values[0] == 1.0)
assert (rad_df[(rad_df.peak_i == 2) & (rad_df.channel_i == 0) &
(rad_df.cycle_i == 0)].signal.values[0] == 5.0)
def peaks(self, fields=None, n_peaks_subsample=None):
df = self._load_df_prop_from_fields(
"peak_df", field_iz=self._fields_to_field_iz(fields))
check.df_t(df, self.peak_df_schema)
if self._has_prop("peak_fit_df"):
fit_df = self._load_df_prop_from_fields(
"peak_fit_df", field_iz=self._fields_to_field_iz(fields))
check.df_t(df, self.peak_fit_df_schema)
df = df.set_index(["field_i", "field_peak_i"]).join(
fit_df.set_index(["field_i", "field_peak_i"]))
# The peaks have a local frame_peak_i but they
# don't have their pan-field peak_i set yet.
df = df.reset_index(drop=True)
df.peak_i = df.index
if n_peaks_subsample is not None:
df = df.sample(n_peaks_subsample, replace=True)
return df
def it_mask_rects():
rects_df = res.mask_rects()
check.df_t(rects_df, SigprocV1Result.mask_rects_df_schema)
assert len(rects_df) == 2 * 2 * 3
def _do_ptm_permutations(df, n_ptms_limit):
"""
df is a dataframe with a single pep_i, and pro_ptm_locs, and pep_offset_in_pro
and aa for each location in the peptide
"""
check.df_t(
df,
dict(pep_i=int, pro_ptm_locs=object, pep_offset_in_pro=int, aa=object),
allow_extra_columns=True,
)
# pro_ptm_locs is identical for all rows
pro_ptm_locs = df.pro_ptm_locs.values[0]
if not pro_ptm_locs:
return []
# get 0-based indices from string representation; these are for the
# entire protein.
ptm_locs_zero_based = [(int(x) - 1) for x in pro_ptm_locs.split(";")]
# get the ptms that coincide with the range spanned by this peptide.
min_pos = df.pep_offset_in_pro.min()
max_pos = df.pep_offset_in_pro.max()
ptm_locs_zero_based = [
x for x in ptm_locs_zero_based if min_pos <= x <= max_pos
]
n_ptms = len(ptm_locs_zero_based)
if n_ptms > n_ptms_limit:
_info(
f"Skipping ptm for peptide {df.pep_i.iloc[0]} with {n_ptms} PTMs")
if n_ptms_limit is not None and n_ptms > n_ptms_limit:
return []
powerset = [
np.array(x) for length in range(1,
len(ptm_locs_zero_based) + 1)
for x in itertools.combinations(ptm_locs_zero_based, length)
]
# powerset is a list of tuples. So if have [2,4,10] in ptms you get
# powerset = [ (2), (4), (10), (2,4), (2,10), (4,10), (2,4,10) ]
#
# The goal is to make a new peptide+seq for each of those tuples by
# adding the modification '[p]' to the aa at that seq index location
#
mod = "[p]"
new_pep_seqs = []
for ptm_locs in powerset:
new_pep_seq = df.copy()
new_pep_seq.pep_i = np.nan
new_pep_seq = new_pep_seq.set_index("pep_offset_in_pro")
new_pep_seq.at[ptm_locs, "aa"] = new_pep_seq.aa[ptm_locs] + mod
new_pep_seq = new_pep_seq.reset_index()
new_pep_seqs += [new_pep_seq]
return new_pep_seqs
def mask_rects(self):
df = self._load_df_prop_from_fields("mask_rects_df")
check.df_t(df, self.mask_rects_df_schema)
return df
def fields(self, fields=None):
df = self._load_df_prop_from_fields(
"field_df", field_iz=self._fields_to_field_iz(fields))
check.df_t(df, self.field_df_schema, allow_extra_columns=True)
return df
def fields(self):
df = self._load_df_prop_from_all_fields("field_df")
check.df_t(df, self.field_df_schema, allow_extra_columns=True)
return df
