Welcome to the phenotrex documentation!

phenotrex

PyPI Codecov Code Quality Travis CI AppVeyor CI Documentation Status

Microbial Phenotype Prediction, re-implemented with Python 3.7 and scikit-learn

  • Supported platforms: Linux, MacOS, Windows

  • Free software: MIT license

Installation

Stable release

To install pheno-trex, run this command in your terminal:

$ pip install phenotrex

This is the preferred method to install pheno-trex, as it will always install the most recent stable release.

If you don’t have pip installed, this Python installation guide can guide you through the process.

From sources

The sources for pheno-trex can be downloaded from the Github repo.

You can either clone the public repository:

$ git clone git://github.com/univieCUBE/PICA2

Or download the tarball:

$ curl  -OL https://github.com/univieCUBE/PICA2/tarball/master

Once you have a copy of the source, you can install it with:

$ python setup.py install

Usage

To use pheno-trex in a project:

from phenotrex.io import ...  #  file I/O
from phenotrex.ml import ...  # classifiers and training/CV functionality
from phenotrex.util import ...  # plotting and util functions

phenotrex

phenotrex package

Subpackages

phenotrex.cli package
Submodules
phenotrex.cli.cccv module
phenotrex.cli.clf_opt module
phenotrex.cli.compute_genotype module
phenotrex.cli.cv module
phenotrex.cli.generic_func module
phenotrex.cli.generic_opt module
phenotrex.cli.get_weights module
phenotrex.cli.main module
phenotrex.cli.plot module
phenotrex.cli.predict module
phenotrex.cli.train module
Module contents
phenotrex.io package
Submodules
phenotrex.io.flat module
phenotrex.io.serialization module
Module contents
phenotrex.ml package
Subpackages
phenotrex.ml.clf package
Submodules
phenotrex.ml.clf.svm module
phenotrex.ml.clf.xgbm module
Module contents
Submodules
phenotrex.ml.cccv module
phenotrex.ml.feature_select module
phenotrex.ml.trex_classifier module
phenotrex.ml.vectorizer module
Module contents
phenotrex.structure package
Submodules
phenotrex.structure.records module
class phenotrex.structure.records.GenotypeRecord(identifier: str, features: List[str])[source]

Bases: object

Genomic features of a sample referenced by identifier.

features: List[str] = None
identifier: str = None
class phenotrex.structure.records.GroupRecord(identifier: str, group_name: Optional[str], group_id: Optional[int])[source]

Bases: object

Group label of sample identifier. Notes —– Useful for leave-one-group-out cross-validation (LOGO-CV), for example, to take taxonomy into account.

group_id: Optional[int] = None
group_name: Optional[str] = None
identifier: str = None
class phenotrex.structure.records.PhenotypeRecord(identifier: str, trait_name: str, trait_sign: int)[source]

Bases: object

Ground truth labels of sample identifier, indicating presence/absence of trait trait_name:

  • 0 if trait is absent

  • 1 if trait is present

identifier: str = None
trait_name: str = None
trait_sign: int = None
class phenotrex.structure.records.TrainingRecord(identifier: str, group_name: Optional[str], group_id: Optional[int], trait_name: str, trait_sign: int, features: List[str])[source]

Bases: phenotrex.structure.records.GenotypeRecord, phenotrex.structure.records.PhenotypeRecord, phenotrex.structure.records.GroupRecord

Sample containing Genotype-, Phenotype- and GroupRecords, suitable as machine learning input for a single observation.

features = None
identifier = None
Module contents
phenotrex.transforms package
Submodules
phenotrex.transforms.annotation module
phenotrex.transforms.resampling module
class phenotrex.transforms.resampling.TrainingRecordResampler(random_state: float = None, verb: bool = False)[source]

Bases: object

Instantiates an object which can generate versions of a TrainingRecord resampled to defined completeness and contamination levels. Requires prior fitting with full List[TrainingRecord] to get sources of contamination for both classes.

Parameters

  • random_state – Randomness seed to use while resampling

  • verb – Toggle verbosity

fit(records: List[phenotrex.structure.records.TrainingRecord])[source]

Fit TrainingRecordResampler on full TrainingRecord list to determine set of positive and negative features for contamination resampling.

Parameters

records – the full List[TrainingRecord] on which ml training will commence.

Returns

True if fitting was performed, else False.

get_resampled(record: phenotrex.structure.records.TrainingRecord, comple: float = 1, conta: float = 0) → phenotrex.structure.records.TrainingRecord[source]

Resample a TrainingRecord to defined completeness and contamination levels. Comple=1, Conta=1 will double set size.

Parameters

  • comple – completeness of returned TrainingRecord features. Range: 0 - 1

  • conta – contamination of returned TrainingRecord features. Range: 0 - 1

  • record – the input TrainingRecord

Returns

a resampled TrainingRecord.

Module contents
phenotrex.transforms.fastas_to_grs(*args, **kwargs)
phenotrex.util package
Submodules
phenotrex.util.helpers module
phenotrex.util.helpers.fail_missing_dependency(*args, **kwargs)[source]
phenotrex.util.helpers.get_groups(records: List[phenotrex.structure.records.TrainingRecord]) → numpy.ndarray[source]

Get groups from list of TrainingRecords

Parameters

records –

Returns

list for groups

phenotrex.util.helpers.get_x_y_tn(records: List[phenotrex.structure.records.TrainingRecord]) → Tuple[numpy.ndarray, numpy.ndarray, str][source]

Get separate X and y from list of TrainingRecord. Also infer trait name from first TrainingRecord.

Parameters

records – a List[TrainingRecord]

Returns

separate lists of features and targets, and the trait name

phenotrex.util.logging module
phenotrex.util.logging.get_logger(initname, verb=False)[source]

This function provides a logger to all scripts used in this project.

Parameters

  • initname – The name of the logger to show up in log.

  • verb – Toggle verbosity

Returns

the finished Logger object.

phenotrex.util.plotting module
phenotrex.util.plotting.compleconta_plot(cccv_results: Union[Dict[float, Dict[float, Dict[str, float]]], List[Dict[float, Dict[float, Dict[str, float]]]]], conditions: List[str] = (), each_n: List[int] = None, title: str = '', fontsize: int = 16, figsize=(10, 7), plot_comple: bool = True, plot_conta: bool = True, colors: List = None, save_path: Union[str, pathlib.Path] = None, **kwargs)[source]

Plots Compleconta CV result for one or multiple models. For perfect completeness and variable contamination as well as perfect contamination and variable completeness, the resulting mean balanced accuracy over folds is plotted.

Parameters

  • cccv_results – a ComplecontaCV result, or list thereof

  • conditions – A list of condition names associated cccv_results

  • each_n – A list of sample counts in datasets associated with cccv_results

  • title – The plot title

  • fontsize – The fontsize of the plot

  • figsize – The figure size (tuple of width, height)

  • plot_comple – Whether to plot completeness

  • plot_conta – Whether to plot contamination

  • colors –

  • save_path – The save path of the plot; if None, display it with plt.show()

  • kwargs – any further keyword arguments passed to plt.plot()

Returns

None

phenotrex.util.taxonomy module
Module contents

Module contents

Top-level package for phenotrex.

Credits

Development Lead

Contributors

History

Indices and tables