Notebook setup

from os import environ
from pathlib import Path
import numpy as np
import pandas as pd
import GSForge as gsf
rng = np.random.default_rng(0)

Creating Feature Sets and Collections¶

A GeneSet is one of the three core data structures provided by GSForge, it stores the result of a selection method so long as those results can be indexed by gene. Any number of measures can be stored, but the support attribute is a special boolean array that indicates ‘selection’ by a given GeneSet.

A GeneSetCollection is the final core data structure, it stores one AnnotatedGEM and any number of GeneSet objects.

In this example we have an AnnotatedGEM already constructed:

OSF_PATH = Path(environ.get("GSFORGE_DEMO_DATA", default="~/GSForge_demo_data/")).expanduser().joinpath("osfstorage", "oryza_sativa")
GEM_PATH = OSF_PATH.joinpath("AnnotatedGEMs", "oryza_sativa_hisat2_raw.nc")

agem = gsf.AnnotatedGEM(GEM_PATH)
agem

<GSForge.AnnotatedGEM>
Name: Oryza Sativa
Selected GEM Variable: 'counts'
    Gene   66338
    Sample 475

Creating GeneSets¶

See the API reference of GSForge.GeneSet for all availble creation functions, which include:

from_pandas
from_GeneSets
from_bool_array
from_gene_array
from_xarray_dataset
from_netcdf

GeneSets from Lists or Arrays¶

A minimal GeneSet is just that: a set of genes.

Here we draw random features to demonstrate this:

random_features = rng.choice(agem.data.Gene.values, 10, replace=False)
random_features

array(['LOC_Os10g39450.1', 'LOC_Os10g12660.1', 'LOC_Os07g34630.1',
       'LOC_Os06g05610.1', 'LOC_Os03g28010.1', 'LOC_Os01g26410.1',
       'LOC_Os01g10420.1', 'LOC_Os03g53690.1', 'LOC_Os02g34160.1',
       'LOC_Os01g49154.1'], dtype=object)

Provide this to the from_gene_array() constructor to create a simple GeneSet.

set_example = gsf.GeneSet.from_gene_array(random_features, name='Random Set')
set_example

<GSForge.GeneSet>
Name: Random Set
    Supported Genes:  10

From `pandas.DataFrames` or `xarray.DataSet` objects¶

Commonly there is some information associated with a set of features. Differential gene expression results often contain information about many (or all) of the genes, but only identify a few as ‘differentially expressed’. We can store all such information in the GeneSet object, and indicate which genes are selected by setting a boolean array named support.

Here we simulate an example DataFrame.

n = 100
random_features = rng.choice(agem.data.Gene.values, n, replace=False)

df = pd.DataFrame(
    {
        'sim_LFC': rng.normal(size=n),
        'sim_pvalue': np.abs(rng.normal(size=n)),
    },
    index=random_features
)

df['support'] = (df['sim_pvalue'] < 0.05) | (df['sim_LFC'] > 1.0)

df.head()

	sim_LFC	sim_pvalue	support
LOC_Os12g34470.1	0.621018	1.292646	False
LOC_Os10g42500.1	-2.250141	0.471813	False
LOC_Os04g50110.1	0.386370	1.377951	False
LOC_Os09g37520.1	-0.581641	0.135731	False
LOC_Os05g26610.1	0.109280	2.310363	False

Create the GeneSet using the from_pandas() constructor.

dge_gs = gsf.GeneSet.from_pandas(df, name='Sim DGE')
dge_gs

<GSForge.GeneSet>
Name: Sim DGE
    Supported Genes:  19

Creating and Saving GeneSetCollections¶

We only need to provided an AnnotatedGEM and a name to create a GeneSetCollection. Then add GeneSet objects like you would entries to a dictionary:

sample_coll = gsf.GeneSetCollection(gem=agem, name='Literature DGE')
sample_coll['set example'] = dge_gs
sample_coll['simulated DGE example'] = set_example
sample_coll

<GSForge.GeneSetCollection>
Literature DGE
GeneSets (2 total): Support Count
    set example: 19
    simulated DGE example: 10

GeneSetCollections are saved as a directory, each set saved as a separate netcdf file.

save = False
if save == True:
    sample_coll.save('my_collection')

Creating Feature Sets and Collections¶

Creating GeneSets¶

GeneSets from Lists or Arrays¶

From pandas.DataFrames or xarray.DataSet objects¶

Creating and Saving GeneSetCollections¶

From `pandas.DataFrames` or `xarray.DataSet` objects¶