def main(env, args):
# type: (Environment, argparse.Namespace) -> None
df_bac = load_obj(args.pf_input_bac) # type: pd.DataFrame
df_arc = load_obj(args.pf_input_arc) # type: pd.DataFrame
df_bac["Type"] = "Bacteria"
df_arc["Type"] = "Archaea"
df = pd.concat([df_bac, df_arc], sort=False)
# df = df.sample(100)
df["GENOME_TYPE"] = df["GENOME_TYPE"].apply(
lambda x: x.strip().split("-")[1].upper())
df.loc[df["GENOME_TYPE"] == "D2", "GENOME_TYPE"] = "D"
df.reset_index(inplace=True)
import matplotlib
matplotlib.rcParams.update({
# "pgf.texsystem": "pdflatex",
'font.family': 'serif',
'text.usetex': False,
'pgf.rcfonts': False,
})
visualize_matrix_column(env, df, "RBS_MAT")
visualize_matrix_column(
env, df[(df["Type"] == "Bacteria") & (df["GENOME_TYPE"] == "C")],
"PROMOTER_MAT")
def main(env, args):
# type: (Environment, argparse.Namespace) -> None
df = load_obj(args.pf_data) # type: pd.DataFrame
df.reset_index(inplace=True)
df = df[df["GENOME_TYPE"] == "group-a"].copy()
df["RE"] = df[["RBS_MAT", "NON_MAT"]].apply(lambda r: relative_entropy(
MotifModel(r["RBS_MAT"], None), GMS2Noncoding(r["NON_MAT"])),
axis=1)
sns.jointplot(df, "GC", "RE")
sns.kdeplot(df, "GC", "RE")
def main(env, args):
# type: (Environment, argparse.Namespace) -> None
mgm_models = load_obj(
args.pf_mgm_models
) # type: Dict[str, Dict[str, Dict[str, MGMMotifModelAllGC]]]
df_test = pd.read_csv(args.pf_test) # type: pd.DataFrame
# df_test = df_test.head(500).copy()
run_mgm_models_on_test_data(env, mgm_models, df_test, args.species_type,
args.pf_output)
# df_test = parallelize_dataframe_by_chunks(df_test, run_mgm_models_on_test_data, "df_test", {
# "env": env, "mgm_models": mgm_models, "species_type": args.species_type, "pf_output": args.pf_output
# })
# return
df_test.to_csv(args.pf_output, index=False)
def main(env, args):
# type: (Environment, argparse.Namespace) -> None
if args.pf_load_state is None:
gcfid_to_number_of_targets = count_targets_per_gcfid(
args.pf_sbsp_output)
df_assembly_summary = read_assembly_summary_into_dataframe(
args.pf_assembly_summary)
gcfid_to_assembly_info = get_assembly_info_per_gcfid(
df_assembly_summary)
taxid_to_number_of_targets = {
int(gcfid_to_assembly_info[gcfid]["taxid"]):
gcfid_to_number_of_targets[gcfid]
for gcfid in gcfid_to_number_of_targets
if gcfid in gcfid_to_assembly_info
}
tree = TaxonomyTree.load(args.pf_taxonomy_tree)
tree.update_tree_attributes(
set_number_of_targets_per_taxid,
{"taxid_to_number_of_targets": taxid_to_number_of_targets},
direction="bottom-up")
if args.pf_save_state is not None:
save_obj(tree, args.pf_save_state)
else:
tree = load_obj(args.pf_load_state)
tree_string = tree.to_string(check_if_should_print=should_print,
attribute_name="number_of_targets",
attribute_format="{:,}",
tag_name=args.tag,
max_depth=args.max_depth)
write_string_to_file(tree_string, args.pf_output)
def load(pf_load):
# type: (str) -> TaxonomyTree
return load_obj(pf_load)
def main(env, args):
# type: (Environment, argparse.Namespace) -> None
df_bac = load_obj(args.pf_data).reset_index() # type: pd.DataFrame
df_bac = df_bac[df_bac["GENOME_TYPE"].isin(args.group)]
min_gc = 20
max_gc = 70
if args.motif_type == "PROMOTER":
df_bac = df_bac[df_bac["GC"] >= 40].copy()
gc_values = np.arange(min_gc, max_gc, 2)
models = get_models_by_gc(df_bac, gc_values, motif_type=args.motif_type)
num_plots = len(models)
num_rows = int(math.sqrt(num_plots))
num_cols = math.ceil(num_plots / float(num_rows))
fig, axes = plt.subplots(num_rows,
num_cols,
sharex="all",
sharey="all",
figsize=(12, 10))
model_index = 0
for r in range(num_rows):
for c in range(num_cols):
if model_index >= len(models):
break
if models[model_index] is None:
model_index += 1
continue
bgd = [0.25] * 4
bgd = background_from_gc(gc_values[model_index])
newmod = lm.transform_matrix(models[model_index][0],
to_type="information",
from_type="probability",
background=models[model_index][1])
# from copy import copy
# newmod = copy(models[model_index][0])
# for idx in newmod.index:
# # see https://bioconductor.org/packages/release/bioc/vignettes/universalmotif/inst/doc/IntroductionToSequenceMotifs.pdf
#
# uncertainty = sum(
# [newmod.at[idx, l] * math.log2(newmod.at[idx, l]) for l in newmod.columns]
# )
# fIC = math.log2(4) - uncertainty
# for i, l in enumerate(sorted(newmod.columns)):
# newmod.at[idx, l] = max(1 * newmod.at[idx, l] * math.log2(newmod.at[idx, l] / models[model_index][1][i]), 0)
lm.Logo(newmod, ax=axes[r][c])
axes[r][c].set_ylim(0, 2)
axes[r][c].set_title(int(gc_values[model_index]))
# fig.show()
model_index += 1
plt.tight_layout()
plt.savefig(next_name(env["pd-work"]))
plt.show()