encodeproject 1.0.22

Queries

The library currently offers to query methods that already integrate some filtering properties: one for the experiments and one for the biosamples.

For querying the experiments you can run the following:

from encodeproject import experiment

experiments = experiment()

Let’s take a look to an in-depth example, showing all the available parameters:

from encodeproject import experiment

experiments = experiment(
    # The cell line we are interested in.
    # For example values could be K562 or GM12878.
    # We use None to specify that we are not
    # interested in any particular cell line.
    cell_line = None,
    # The reference genomic assembly we want.
    # For example values could be hg19 or GRCh38
    # We use None to specify that we are not
    # interested in any particular genomic assembly.
    assembly = None,
    # The target (the genes coding for proteins in this context) we want.
    # For example values could be CTCF or H3K27ac
    # We use None to specify that we are not
    # interested in any particular target.
    target = None,
    # The status of the data we want.
    # We only want released data, meaning data that are
    # neither old (archived) or with errors (revoked).
    status = 'released',
    # The organism we are considering.
    # Since we only want Homo sapiens data,
    # we specify that organism name.
    organism = 'Homo sapiens',
    # The format of the files we are interested in
    file_type = 'bigWig',
    # We ask to consider only experiments with replicas
    replicated = True,
    # We only want with the signals
    # expressed as "fold change over control"
    searchTerm = "fold change over control",
    # We do not need to specify any other specific
    # additional parameters
    parameters = None,
    # We want to download all the
    # available experiments
    limit = 'all',
    # We want to drop all the experiments
    # which have been characterized by significand issues
    drop_errors = (
        'extremely low read depth',
        'missing control alignments',
        'control extremely low read depth',
        'extremely low spot score',
        'extremely low coverage',
        'extremely low read length',
        'inconsistent control',
        'inconsistent read count'
    )
)

All parameters are optional, they just act as additional filters.

For querying the biosamples you can run the following:

from encodeproject import biosample

my_biosample_query_response = biosample(
    accession="ENCSR000EDP", # The accession code for the desired biosample
)

As for the experiments there are a number of filters available:

hg19_samples = biosamples(
    # The list of accessions to retrieve
    accessions=accession_codes,
    # Wethever to convert the results in dataframe.
    # The following filters only apply if dataframes are used
    to_dataframe = True,
    # The status of the data we want.
    # We only want released data, meaning data that are
    # neither old (archived) or with errors (revoked).
    status = "released",
    # The organism we want.
    organism = "human",
    # The genomic assembly we want to use
    assembly = "hg19",
    # The output type we want.
    output_type = "fold change over control",
    # And finally the bare minimum amount
    # of biological replicates
    min_biological_replicates = 2
)

For running multiple queries for biosamples at once you can run the following:

from encodeproject import biosamples

responses = biosamples(
    accessions=["ENCSR000EDP", "ENCSR030EDP", "ENCSR067EDP"], # The accessions code for the desired biosamples
)

Filters

Since the response files can get quite big and hard to read, I’ve prepared also a couple filter functions.

For filtering the accessions codes from an experiment response you can use:

from encodeproject import accessions

codes = accessions(my_experiment_query_response)

For filtering the download URLs from a biosample response you can use:

from encodeproject import download_urls

codes = download_urls(my_biosample_query_response)

Download utility

I’ve added also a method to download from a given URL, showing a loading bar, based on this answer from StackOverflow.

from encodeproject import download

download("https://encode-public.s3.amazonaws.com/2012/07/01/074e1b37-2be1-4f6a-aa42-6c512fd1834b/ENCFF000XOW.bigWig")

Sample to DataFrame instruction

Utility to convert a sample to a relatively simple pandas DataFrame.

from encodeproject import biosample_to_dataframe

df = biosample_to_dataframe(my_biosample_query_response)