🚧 PFB Python SDK 🚧

Python SDK to create, explore and modify PFB (Portable Format for Biomedical Data) files.

PyPFB Overview

For more details, see the Portable Format for Biomedical Data document.

PyPFB is a python sdk to create, explore, and modify PFB (Portable Format for Bioinformatics) files.

These files start from a Gen3 data dictionary. These can be made from either json hosted on a cloud storage service, like s3, or from a local directory. See PyPFB From Schema for an example.

Once we have a PFB file created from a schema we can start to add data to the file. This is done using JSON files from local directory. We create them in the style of our data-simulator https://github.com/uc-cdis/data-simulator/ Once we have them we can use PFB From JSON to import the structured json into our Serialized PFB file.

At this point we have a PFB file with married schema and serialized data. Now we have a few options for modifying these PFB files. These are good options for breaking changes within the dictionary. This allows a commons operator to export the entire structured database, make modifications to fix the breaking changes, and then re import the file back to the commons.

Changes that are already supported by this SDK are renames (enum and nodes) and adds of records data.

PFB Schema

Installation

From PyPI:

pip install pypfb[gen3]

(The optional gen3 dependencies add the ability to convert a Gen3 data dictionary into a PFB file.)

From source code:

poetry install

Usage

Main

Usage: pfb [OPTIONS] COMMAND [ARGS]...

  PFB: Portable Format for Biomedical Data.

Commands:
  add     Add records into a PFB file.
  from    Generate PFB from other data formats.
  make    Make a blank record for add.
  rename  Rename different parts of schema.
  show    Show different parts of a PFB file.
  to      Convert PFB into other data formats.

Show different parts of PFB

Usage: pfb show [OPTIONS] COMMAND [ARGS]...

  Show records of the PFB file.

  Specify a sub-command to show other information.

Options:
  -i, --input FILENAME  The PFB file.  [default: <stdin>]
  -n, --limit INTEGER   How many records to show, ignored for sub-commands.
                        [default: no limit]

Commands:
  metadata  Show the metadata of the PFB file.
  nodes     Show all the node names in the PFB file.
  schema    Show the schema of the PFB file.

Examples:
  schema:
    pfb show -i data.avro schema
  nodes:
    pfb show -i data.avro nodes
  metadata:
    pfb show -i data.avro metadata
  records:
    pfb show -i data.avro -n 5

Convert Gen3 data dictionary into PFB schema

Usage: pfb from [PARENT OPTIONS] dict DICTIONARY

  Convert Gen3 data DICTIONARY into a PFB file.

  If DICTIONARY is a HTTP URL, it will be downloaded and parsed as JSON; or
  it will be treated as a local path to a directory containing YAML files.

Parent Options:
  -o, --output FILENAME  The output PFB file.  [default: <stdout>]

Examples:
  URL:
    pfb from -o thing.avro dict https://s3.amazonaws.com/dictionary-artifacts/gtexdictionary/3.2.2/schema.json
  Directory:
    pfb from -o gdc.avro dict /path/to/dictionary/schemas/

Convert JSON for corresponding datadictionary to PFB

Usage: pfb from [PARENT OPTIONS] json [OPTIONS] [PATH]

  Convert JSON files under PATH into a PFB file.

Parent Options:
  -o, --output FILENAME  The output PFB file.  [default: <stdout>]

Options:
  -s, --schema FILENAME  The PFB file to load the schema from.  [required]
  --program TEXT         Name of the program.  [required]
  --project TEXT         Name of the project.  [required]

Example:
  pfb from -o data.avro json -s schema.avro --program DEV --project test /path/to/data/json/

Convert TSV for corresponding datadictionary to PFB

Usage: pfb from [PARENT OPTIONS] tsv [OPTIONS] [PATH]

  Convert TSV files under PATH into a PFB file.

Parent Options:
  -o, --output FILENAME  The output PFB file.  [default: <stdout>]

Options:
  -s, --schema FILENAME  The PFB file to load the schema from.  [required]
  --program TEXT         Name of the program.  [required]
  --project TEXT         Name of the project.  [required]

Example:
  pfb from -o data.avro tsv -s schema.avro --program DEV --project test /path/to/data/tsv/

Make new blank record

Usage: pfb make [OPTIONS] NAME

  Make a blank record according to given NODE schema in the PFB file.

Options:
  -i, --input PFB  Read schema from this PFB file.  [default: <stdin>]

Example:
  pfb make -i test.avro demographic > empty_demographic.json

Add new record to PFB

Usage: pfb add [OPTIONS] PFB

  Add records from a minified JSON file to the PFB file.

Options:
  -i, --input JSON  The JSON file to add.  [default: <stdin>]

Example:
  pfb add -i new_record.json pfb.avro

Rename different parts of PFB (schema evolution)

Usage: pfb rename [OPTIONS] COMMAND [ARGS]...

  Rename different parts of schema.

Options:
  -i, --input FILENAME   Source PFB file.  [default: <stdin>]
  -o, --output FILENAME  Destination PFB file.  [default: <stdout>]

Commands:
  enum  Rename enum.
  node  Rename node.
  type  Rename type (not implemented).

Examples:
  enum:
    pfb rename -i data.avro -o data_enum.avro enum demographic_ethnicity old_enum new_enum
  node:
    pfb rename -i data.avro -o data_update.avro node demographic information

Rename node

Usage: pfb rename [PARENT OPTIONS] node [OPTIONS] OLD NEW

  Rename node from OLD to NEW.

Rename enum

Usage: pfb rename [PARENT OPTIONS] enum [OPTIONS] FIELD OLD NEW

  Rename enum of FIELD from OLD to NEW.

Convert PFB into Neptune (bulk load format for Gremlin)

Usage: pfb to [PARENT OPTIONS] gremlin [OPTIONS] [OUTPUT]

  Convert PFB into CSV files under OUTPUT for Neptune bulk load (Gremlin).

  The default OUTPUT is ./gremlin/.

Options:
  --gzip / --no-gzip  Whether gzip the output.  [default: yes]

Example:
  pfb to -i data.avro gremlin

Convert PFB into TSV (1 TSV per node)

Usage: pfb to [PARENT OPTIONS] tsv [OPTIONS] [OUTPUT]

  Convert PFB into TSV files under [OUTPUT] for modification of data in TSV format.

  The default [OUTPUT] is ./tsvs/.

Options:
  None
Example:
  pfb to -i data.avro tsv

PFB ETL

Usage: pfb etl [OPTIONS] PFB

  ETL PFB into ES indices

Options:
  -u, --url URL      base es url
  -t, --token TOKEN  access token
  -n, --node NODE    root node for ETL
Example:
  pfb etl -u http://localhost:9200  -n participant ./tests/pfb-data/test.avro

Example of minimal PFB

In the examples/minimal-pfb directory we have an example of a minimal pfb that only contains submitted unaligned read data

First create the PFB with schema from the json dictionary
pfb from -o minimal_schema.avro dict minimal_file.json

Then we put the data into the PFB
pfb from -o minimal_data.avro json -s minimal_schema.avro --program DEV --project test sample_file_json/

We can view the data of the PFB
pfb show -i minimal_data.pfb

We can also view the schema of the PFB
pfb show -i minimal_data.pfb schema

Examples

pfb from dict http://s3.amazonaws.com/dictionary-artifacts/kf-dictionary/1.1.0/schema.json > ./tests/schema/kf.avro

pfb from json ./tests/data -s ./tests/schema/kf.avro --program DEV --project test > tests/pfb-data/test.avro

pfb from tsv ./tests/tsv_data -s ./tests/schema/kf.avro --program DEV --project test > tests/pfb-data/test.avro

cat tests/pfb-data/test.avro | pfb rename node slide slide_test > tests/pfb-data/rename_test.avro

cat tests/pfb-data/test.avro | pfb rename enum state validated validated_test > tests/pfb-data/rename_test.avro

cat tests/pfb-data/test.avro | pfb show -n 1 | jq

cat tests/pfb-data/test.avro | pfb show --schema | jq

cat tests/pfb-data/test.avro | pfb to gremlin ./output/
cat tests/pfb-data/test.avro | pfb to tsv ./tsvs/