Python

Table of Contents

• Installation
• Usage
• Functions

Installation

To install the latest release from PyPI:

pip install bedrock-bio

Usage

Load package:

import bedrock_bio as bb

List available namespaces and tables:

bb.list_namespaces()
bb.list_tables()

Lazily load a table (returns a DuckDBPyRelation object):

pqtls_tbl = bb.load_table('ukb_ppp.pqtls')

Filter columns and rows of lazy table using the DuckDB Relational API:

pqtls_tbl = (pqtls_tbl
    .filter(
        "ancestry = 'EUR'"
    )
    .filter(
        "protein_id = 'A0FGR8'"
    )
    .filter(
        "neg_log_10_p_value >= 7.30"  # p-value ~5e-8
    )
    .select(
        'chromosome',
        'position_grch38',
        'effect_allele',
        'other_allele',
        'beta',
        'neg_log_10_p_value'
    ))

Note

For best performance, filter rows and columns on the lazy DuckDB table before converting to Polars or Pandas DataFrames. This reduces the amount of data downloaded, which speeds up analysis and reduces memory pressure on the local machine.

It is particularly recommended to filter on the partitioning columns of a table. Use describe_table() or check a table’s documentation page to see the partitioning columns and their allowed values.

Convert to Polars DataFrame (triggers data download):

pqtls_df = pqtls_tbl.pl()

Or, convert to Pandas DataFrame (also triggers download):

pqtls_df = pqtls_tbl.df()

Functions

Signature	Description	Returns
list_namespaces()	List namespace identifiers.	list[str]
`list_tables(namespace: str	None = None)`	List all tables, filtered to a namespace if specified.
describe_namespace(name: str)	Inspect namespace metadata.	dict
describe_table(name: str)	Inspect table metadata.	dict
load_table(name: str)	Lazily load the specified table identifier.	DuckDBPyRelation