ont-pyguppy-client-lib 7.2.15

Python bindings for the GuppyClient library.

Important notice

ont-pyguppy-client-lib has been renamed to ont-pybasecall-client-lib. This project will no longer be updated but is provided in order to support basecall servers up to version 7.2.x. For python bindings to dorado_basecall_server 7.3.0 onwards, please use ont-pybasecall-client-lib.

ont-pyguppy-client-lib

ont-pyguppy-client-lib provides python bindings for connecting to a Dorado basecall server. It allows you to interact with the server to do anything you could normally do using the ont_basecall_client. This includes:

• Basecalling

• Barcoding / demultiplexing

• Alignment

For example:

>>> from pyguppy_client_lib.pyclient import PyGuppyClient
>>> client = PyGuppyClient(
    "127.0.0.1:5555",
    "dna_r9.4.1_450bps_fast",
    align_ref="/path/to/index.mmi",
    bed_file="/path/to/targets.bed"
)
>>> client.connect()

Getting started

ont-pyguppy-client-lib is available on PyPI and may be installed via pip:

pip install ont-pyguppy-client-lib

ont-pyguppy-client-lib requires an instance of the Dorado basecall server is running. ont-dorado-server may be obtained from the Oxford Nanopore Community

The version of ont-pyguppy-client-lib should exactly match the version of ont-dorado-server being used. You can find your ont-dorado-server version like this:

$ <location of dorado_basecall_server>/dorado_basecall_server --version

For example, this Dorado basecall server is version 7.1.1:

$ ./ont-dorado-server/bin/dorado_basecall_server --version
: Dorado Basecall Service Software, (C) Oxford Nanopore Technologies,  Limited. Version 7.1.1+effbaf8, client-server API version 16.0.0

Install a specific version of ont-pyguppy-client-lib like this:

pip install ont-pyguppy-client-lib==<version>

Dependencies

ont-pyguppy-client-lib requires numpy in order to run. In order to use included helper functions for reading data from fast5 and/or pod5 files it is also necessary to manually install ont-fast5-api and/or pod5:

pip install ont-fast5-api pod5

Documentation and help

Information on the methods available may be viewed through Python’s help command::

>>> from pyguppy_client_lib import pyclient
>>> help(pyclient)
>>> from pyguppy_client_lib import client_lib
>>> help(client_lib)

Interface / Examples

ont-pyguppy-client-lib comprises three Python modules:

1. pyclient A user-friendly wrapper around client_lib. This is what you should use to interact with a Dorado basecall server.

2. client_lib A compiled library which provides direct Python bindings to Dorado’s C++ GuppyClient API.

3. helper_functions A set of functions for running a Dorado basecall server and loading reads from fast5 and/or pod5 files.

Starting a basecall server

There must be a Dorado basecall server running in order to communicate with it. On most Oxford Nanopore devices a basecall server is always running on port 5555. On other devices, or if you want to run a separate basecall server, you must start one yourself:

from pyguppy_client_lib import helper_functions

# A basecall server requires:
#  * A location to put log files (on your PC)
#  * An initial config file to load
#  * A port to run on
server_args = ["--log_path", "/home/myuser/guppy_server_logs",
               "--config", "dna_r9.4.1_450bps_fast.cfg",
               "--port", 5556]
# The second argument is the directory where the
# dorado_basecall_server executable is found. Update this as
# appropriate.
helper_functions.run_server(server_args, "/home/myuser/ont-dorado/bin")

See the the DOCUMENTATION.md file in the ont-dorado-server archive for more information on server arguments.

Basecall and align using PyGuppyClient

from pyguppy_client_lib.pyclient import PyGuppyClient

client = PyGuppyClient(
    "127.0.0.1:5555",
    "dna_r9.4.1_450bps_fast",
    align_ref = "/path/to/align_ref.fasta",
    bed_file = "/path/to/bed_file.bed"
)
client.connect()

Note that the helper_functions module requires that ont-fast5-api and/or pod5 is installed.:

from pyguppy_client_lib.helper_functions import basecall_with_pyguppy

# Using the client generated in the previous example
called_reads = basecall_with_pyguppy(
    caller,
    "/path/to/input_folder"
)

for read in called_reads:
    read_id = read['metadata']['read_id']
    alignment_genome = read['metadata']['alignment_genome']
    sequence = read['datasets']['sequence']
    print(f"{read_id} sequence length is {len(sequence)}"
          f"alignment_genome is {alignment_genome}")

Basecall and get states, moves and modbases using GuppyClient

In order to retrieve the movement dataset, the move_and_trace_enabled option must be set to True; analogously, for the state_data one, post_out must be turned on. NOTE: You shouldn’t turn on post_out if you don’t need the states, because it generates a LOT of extra output data so it can really hurt performance. Likewise with move_and_trace_enabled, although that’s much less expensive.

options = {'priority': GuppyClient.high_priority,
          'client_name': "test_client",
          'move_and_trace_enabled': True,
          'post_out':True }

client = GuppyClient(port_path, 'dna_r9.4.1_e8.1_modbases_5mc_cg_fast')
result = client.set_params(options)
result = client.connect()

called_reads = basecall_with_pyguppy(client, input_path)

for read in called_reads:
    base_mod_context = read['metadata']['base_mod_context']
    base_mod_alphabet = read['metadata']['base_mod_alphabet']

    sequence = read['datasets']['sequence']
    movement = read['datasets']['movement']
    state_data = read['datasets']['state_data']
    base_mod_probs = read['datasets']['base_mod_probs']

    print(f"{read_id} sequence length is {len(sequence)}, "
          f"base_mod_context is {base_mod_context}, base_mod_alphabet is {base_mod_alphabet}, "
          f"movement size is {movement.shape}, state_data size is {state_data.shape}, "
          f"base_mod_probs size is {base_mod_probs.shape}")

Glossary of Terms:

Dorado - Oxford Nanopore Technologies’ production basecaller, which translates electrical signals measured from nanopores into DNA or RNA bases.

Fast5 - an implementation of the HDF5 file format, with specific data schemas for Oxford Nanopore Technologies sequencing data.

Pod5 - a file format for storing nanopore dna data in an easily accessible way.