def _default_parse_func(file, shape, sep, encoding):
data = []
nr, nc = shape
rxc = nr * nc
with open(file, "r", encoding=encoding) as fh:
for line in fh:
splitted = line.replace("\n", "").split(sep)[2:]
if (len(splitted) == rxc):
data.append(splitted)
if not data:
raise AsRuntimeError(
"""DataParser: parse failed, no any valid line found
make sure data file is in correct format""")
if (len(data) % 2):
raise AsRuntimeError(
"""DataParser: parse failed, uneven GFP and OD sections
make sure data file is in correct format""")
data = numpy.asarray(data, dtype=object)
sz_d1 = len(data) // 2
# replace overflow with 100000
data[data == "OVRFLW"] = 100000
OD = data[1:sz_d1, :].reshape(-1, nr, nc).astype(float)
GFP = data[sz_d1 + 1:].reshape(-1, nr, nc).astype(int)
MASK = (GFP != 100000)
return _PlateData(OD, GFP, MASK)
def analyze(self):
# analyze P
for sample in self.samples:
sample.run_P_analysis()
# analyze I
if (self.untreated_sample() is None):
raise AsRuntimeError(
"cannot canculate I with no assign of untreated sample")
untreated_P = self.untreated_sample().P()
for sample in self.get_samples_except_untreated():
sample.run_XELI_analysis(untreated_P)
def _blank_correction(self):
where_blank = self.layout.mask_by_category(self._blank)
if not where_blank.any():
raise AsRuntimeError(
"background correction requires at least one 'BLANK' category in layout"
)
OD_blank = self.OD()[:, where_blank].mean(axis=1, keepdims=True)
self.OD()[:] = self.OD() - OD_blank
GFP_blank = self.GFP()[:, where_blank].mean(axis=1, keepdims=True)
self.GFP()[:] = self.GFP() - GFP_blank
m = ">%s:BLANK_CORRECTION\n%s\n"
return m % (self.name(), self.cat_OD_GFP_tables())
def _parse_maps(gene_map, cate_map): coords_list, genes_list, cates_list = [], [], [] for row_id, row in enumerate(gene_map): for col_id, gene in enumerate(row): cate = cate_map[row_id][col_id] if (gene and cate): coords_list.append([row_id, col_id]) genes_list.append(gene) cates_list.append(cate) elif not (gene or cate): continue else: e = "'gene' or 'category' missing at (row %d, col %d)" raise AsRuntimeError(e % (row_id, col_id)) coords = numpy.asarray(coords_list, dtype = int) genes = numpy.asarray(genes_list, dtype = str) cates = numpy.asarray(cates_list, dtype = str) return coords, genes, cates
def _eline_correction(self):
eline_where = self.layout.mask_by_category(self._eline)
if not eline_where.any():
raise AsRuntimeError(
"e-line correction requires at least one 'ELINE' category in layout"
)
OD_el = self.OD()[:, eline_where]
GFP_el = self.GFP()[:, eline_where]
MASK_el = self.MASK()[:, eline_where]
# use a linear model for eline correction
# the background gfp signal is estimated to be (slope * OD + intercept)
slope, inter, inter_sd, msg = self.eline_corre.feed(OD_el, GFP_el, MASK_el)
self.model_slope = slope
self.model_inter = inter
self.model_inter_sd = inter_sd
m = ">%s:ELINE_CORRECTION\n%s\n"
return m % (self.name(), msg)
def select_slope_and_intercept(self, nr, nc, all_regs):
# select slope and intercept by interactive selector
# however if only one, no need for interactive select
# you have to use that
n = len(all_regs)
if n == 1:
slope_i, inter_i = [0], [0]
else:
slope_i, inter_i = self.selector.launch(n, nr, nc, self.plot_path,
self.sample_name)
if not slope_i:
raise NoValueSelectedError("slopes cannot be null selection")
if not inter_i:
raise NoValueSelectedError("intercepts cannot be null selection")
slopes = all_regs[slope_i, 0]
inters = all_regs[inter_i, 1]
if len(inters) == 1:
raise AsRuntimeError(
"must be at least 2 intercepts (this exception is thrown due to not implemented yet)"
)
slope = slopes.mean()
inter = inters.mean()
inter_sd = inters.std()
ret_msg = """Raw regressions:
Slope\tIntercept\tr^2\tp\tStd.err
%s
Selected:
Slopes: %s
Intercepts: %s
Final:
Slope: %f
Intercept: %f
Inter.SD: %f""" % (array2d2string_by_row(
all_regs.tolist(), "%f\t%f\t%f\t%.8e\t%f"), vector2string(slopes),
vector2string(inters), slope, inter, inter_sd)
return slope, inter, inter_sd, ret_msg
def add_sample(self,
SampleClass,
name,
layout=None,
offset=None,
untreated=False,
**kw):
# internal _id is same as the position in the self.samples array
_id = len(self.samples)
sample = SampleClass(name=name,
layout=layout or self.shared_layout,
log=self.log(),
assay_data=self.data(),
outdir=self.output_dir(),
offset=offset,
_id=len(self.samples),
**kw)
self.samples.append(sample)
if untreated:
if not (self._control is None):
raise AsRuntimeError(
"assign more than one sample as untreated is not allowed")
self._control = sample
def create_output_dir(self, outdir, overwrite=False, **kw):
if os.path.exists(outdir):
if not overwrite:
raise AsRuntimeError("output dir '%s' already exists" % outdir)
else:
os.makedirs(outdir)
def _check_shape(f1, f2):
if (len(f1) != len(f2)):
raise AsRuntimeError("map files must be in exact same shape")
for s1, s2 in zip(f1, f2):
if (len(s1) != len(s2)):
raise AsRuntimeError("map files must be in exact same shape")