diff options
Diffstat (limited to 'gn3')
| -rw-r--r-- | gn3/db/traits.py | 93 | ||||
| -rw-r--r-- | gn3/heatmaps.py | 67 | ||||
| -rw-r--r-- | gn3/partial_correlations.py | 88 |
3 files changed, 182 insertions, 66 deletions
diff --git a/gn3/db/traits.py b/gn3/db/traits.py index f2673c8..1c6aaa7 100644 --- a/gn3/db/traits.py +++ b/gn3/db/traits.py @@ -1,12 +1,81 @@ """This class contains functions relating to trait data manipulation""" import os +from functools import reduce from typing import Any, Dict, Union, Sequence + from gn3.settings import TMPDIR from gn3.random import random_string from gn3.function_helpers import compose from gn3.db.datasets import retrieve_trait_dataset +def export_trait_data( + trait_data: dict, samplelist: Sequence[str], dtype: str = "val", + var_exists: bool = False, n_exists: bool = False): + """ + Export data according to `samplelist`. Mostly used in calculating + correlations. + + DESCRIPTION: + Migrated from + https://github.com/genenetwork/genenetwork1/blob/master/web/webqtl/base/webqtlTrait.py#L166-L211 + + PARAMETERS + trait: (dict) + The dictionary of key-value pairs representing a trait + samplelist: (list) + A list of sample names + dtype: (str) + ... verify what this is ... + var_exists: (bool) + A flag indicating existence of variance + n_exists: (bool) + A flag indicating existence of ndata + """ + def __export_all_types(tdata, sample): + sample_data = [] + if tdata[sample]["value"]: + sample_data.append(tdata[sample]["value"]) + if var_exists: + if tdata[sample]["variance"]: + sample_data.append(tdata[sample]["variance"]) + else: + sample_data.append(None) + if n_exists: + if tdata[sample]["ndata"]: + sample_data.append(tdata[sample]["ndata"]) + else: + sample_data.append(None) + else: + if var_exists and n_exists: + sample_data += [None, None, None] + elif var_exists or n_exists: + sample_data += [None, None] + else: + sample_data.append(None) + + return tuple(sample_data) + + def __exporter(accumulator, sample): + # pylint: disable=[R0911] + if sample in trait_data["data"]: + if dtype == "val": + return accumulator + (trait_data["data"][sample]["value"], ) + if dtype == "var": + return accumulator + (trait_data["data"][sample]["variance"], ) + if dtype == "N": + return accumulator + (trait_data["data"][sample]["ndata"], ) + if dtype == "all": + return accumulator + __export_all_types(trait_data["data"], sample) + raise KeyError("Type `%s` is incorrect" % dtype) + if var_exists and n_exists: + return accumulator + (None, None, None) + if var_exists or n_exists: + return accumulator + (None, None) + return accumulator + (None,) + + return reduce(__exporter, samplelist, tuple()) + def get_trait_csv_sample_data(conn: Any, trait_name: int, phenotype_id: int): """Fetch a trait and return it as a csv string""" @@ -674,3 +743,27 @@ def generate_traits_filename(base_path: str = TMPDIR): """Generate a unique filename for use with generated traits files.""" return "{}/traits_test_file_{}.txt".format( os.path.abspath(base_path), random_string(10)) + +def export_informative(trait_data: dict, inc_var: bool = False) -> tuple: + """ + Export informative strain + + This is a migration of the `exportInformative` function in + web/webqtl/base/webqtlTrait.py module in GeneNetwork1. + + There is a chance that the original implementation has a bug, especially + dealing with the `inc_var` value. It the `inc_var` value is meant to control + the inclusion of the `variance` value, then the current implementation, and + that one in GN1 have a bug. + """ + def __exporter__(acc, data_item): + if not inc_var or data_item["variance"] is not None: + return ( + acc[0] + (data_item["sample_name"],), + acc[1] + (data_item["value"],), + acc[2] + (data_item["variance"],)) + return acc + return reduce( + __exporter__, + filter(lambda td: td["value"] is not None, trait_data["data"].values()), + (tuple(), tuple(), tuple())) diff --git a/gn3/heatmaps.py b/gn3/heatmaps.py index 2dd9d07..bf9dfd1 100644 --- a/gn3/heatmaps.py +++ b/gn3/heatmaps.py @@ -14,6 +14,7 @@ from plotly.subplots import make_subplots # type: ignore from gn3.settings import TMPDIR from gn3.random import random_string from gn3.computations.slink import slink +from gn3.db.traits import export_trait_data from gn3.computations.correlations2 import compute_correlation from gn3.db.genotypes import ( build_genotype_file, load_genotype_samples) @@ -26,72 +27,6 @@ from gn3.computations.qtlreaper import ( parse_reaper_main_results, organise_reaper_main_results) -def export_trait_data( - trait_data: dict, samplelist: Sequence[str], dtype: str = "val", - var_exists: bool = False, n_exists: bool = False): - """ - Export data according to `samplelist`. Mostly used in calculating - correlations. - - DESCRIPTION: - Migrated from - https://github.com/genenetwork/genenetwork1/blob/master/web/webqtl/base/webqtlTrait.py#L166-L211 - - PARAMETERS - trait: (dict) - The dictionary of key-value pairs representing a trait - samplelist: (list) - A list of sample names - dtype: (str) - ... verify what this is ... - var_exists: (bool) - A flag indicating existence of variance - n_exists: (bool) - A flag indicating existence of ndata - """ - def __export_all_types(tdata, sample): - sample_data = [] - if tdata[sample]["value"]: - sample_data.append(tdata[sample]["value"]) - if var_exists: - if tdata[sample]["variance"]: - sample_data.append(tdata[sample]["variance"]) - else: - sample_data.append(None) - if n_exists: - if tdata[sample]["ndata"]: - sample_data.append(tdata[sample]["ndata"]) - else: - sample_data.append(None) - else: - if var_exists and n_exists: - sample_data += [None, None, None] - elif var_exists or n_exists: - sample_data += [None, None] - else: - sample_data.append(None) - - return tuple(sample_data) - - def __exporter(accumulator, sample): - # pylint: disable=[R0911] - if sample in trait_data["data"]: - if dtype == "val": - return accumulator + (trait_data["data"][sample]["value"], ) - if dtype == "var": - return accumulator + (trait_data["data"][sample]["variance"], ) - if dtype == "N": - return accumulator + (trait_data["data"][sample]["ndata"], ) - if dtype == "all": - return accumulator + __export_all_types(trait_data["data"], sample) - raise KeyError("Type `%s` is incorrect" % dtype) - if var_exists and n_exists: - return accumulator + (None, None, None) - if var_exists or n_exists: - return accumulator + (None, None) - return accumulator + (None,) - - return reduce(__exporter, samplelist, tuple()) def trait_display_name(trait: Dict): """ diff --git a/gn3/partial_correlations.py b/gn3/partial_correlations.py new file mode 100644 index 0000000..c556d10 --- /dev/null +++ b/gn3/partial_correlations.py @@ -0,0 +1,88 @@ +""" +This module deals with partial correlations. + +It is an attempt to migrate over the partial correlations feature from +GeneNetwork1. +""" + +from functools import reduce +from typing import Any, Sequence + +def control_samples(controls: Sequence[dict], sampleslist: Sequence[str]): + """ + Fetches data for the control traits. + + This migrates `web/webqtl/correlation/correlationFunction.controlStrain` in + GN1, with a few modifications to the arguments passed in. + + PARAMETERS: + controls: A map of sample names to trait data. Equivalent to the `cvals` + value in the corresponding source function in GN1. + sampleslist: A list of samples. Equivalent to `strainlst` in the + corresponding source function in GN1 + """ + def __process_control__(trait_data): + def __process_sample__(acc, sample): + if sample in trait_data["data"].keys(): + sample_item = trait_data["data"][sample] + val = sample_item["value"] + if val is not None: + return ( + acc[0] + (sample,), + acc[1] + (val,), + acc[2] + (sample_item["variance"],)) + return acc + return reduce( + __process_sample__, sampleslist, (tuple(), tuple(), tuple())) + + return reduce( + lambda acc, item: ( + acc[0] + (item[0],), + acc[1] + (item[1],), + acc[2] + (item[2],), + acc[3] + (len(item[0]),), + ), + [__process_control__(trait_data) for trait_data in controls], + (tuple(), tuple(), tuple(), tuple())) + +def dictify_by_samples(samples_vals_vars: Sequence[Sequence]) -> Sequence[dict]: + """ + Build a sequence of dictionaries from a sequence of separate sequences of + samples, values and variances. + + This is a partial migration of + `web.webqtl.correlation.correlationFunction.fixStrains` function in GN1. + This implementation extracts code that will find common use, and that will + find use in more than one place. + """ + return tuple( + { + sample: {"sample_name": sample, "value": val, "variance": var} + for sample, val, var in zip(*trait_line) + } for trait_line in zip(*(samples_vals_vars[0:3]))) + +def fix_samples(primary_trait: dict, control_traits: Sequence[dict]) -> Sequence[Sequence[Any]]: + """ + Corrects sample_names, values and variance such that they all contain only + those samples that are common to the reference trait and all control traits. + + This is a partial migration of the + `web.webqtl.correlation.correlationFunction.fixStrain` function in GN1. + """ + primary_samples = tuple( + present[0] for present in + ((sample, all(sample in control.keys() for control in control_traits)) + for sample in primary_trait.keys()) + if present[1]) + control_vals_vars: tuple = reduce( + lambda acc, x: (acc[0] + (x[0],), acc[1] + (x[1],)), + ((item["value"], item["variance"]) + for sublist in [tuple(control.values()) for control in control_traits] + for item in sublist), + (tuple(), tuple())) + return ( + primary_samples, + tuple(primary_trait[sample]["value"] for sample in primary_samples), + control_vals_vars[0], + tuple(primary_trait[sample]["variance"] for sample in primary_samples), + control_vals_vars[1]) |