nucleus.io.genomics_writer -- Classes that provide the interface for writing genomics data.

Source code: nucleus/io/genomics_writer.py

Documentation index: doc_index.md

---

GenomicsWriter defines the core API supported by writers, and is subclassed directly or indirectly (via DispatchingGenomicsWriter) for all concrete implementations.

TFRecordWriter is an implementation of the GenomicsWriter API for reading TFRecord files. This is usable for all data types when writing data as serialized protocol buffers.

DispatchingGenomicsWriter is an abstract class defined for convenience on top of GenomicsWriter that supports writing to either the native file format or to TFRecord files of the corresponding protocol buffer used to encode data of that file type. The output format chosen is dependent upon the filename to which the data are being written.

Concrete implementations for individual file types (e.g. BED, SAM, VCF, etc.) reside in type-specific modules in this package. A general example of the write functionality is shown below.

# options is a writer-specific set of options.
options = ...

# records is an iterable of protocol buffers of the specific data type.
records = ...

with GenomicsWriterSubClass(output_path, options) as writer:
  for proto in records:
    writer.write(proto)

Classes overview

Name	Description
DispatchingGenomicsWriter	A GenomicsWriter that dispatches based on the file extension.
GenomicsWriter	Abstract base class for writing genomics data.
TFRecordWriter	A GenomicsWriter that writes to a TFRecord file.

Classes

DispatchingGenomicsWriter

A GenomicsWriter that dispatches based on the file extension.

If '.tfrecord' is present in the filename, a TFRecordWriter is used.
Otherwise, a native writer is.

Sub-classes of DispatchingGenomicsWriter must define a _native_writer()
method.

Methods:

__init__(self, output_path, **kwargs)

Initializer.

Args:
  output_path: str. The output path to which the records are written.
  **kwargs: k=v named args. Keyword arguments used to instantiate the native
    writer, if applicable.

write(self, proto)

GenomicsWriter

Abstract base class for writing genomics data.

A GenomicsWriter only has one method, write, which writes a single
protocol buffer to a file.

Methods:

write(self, proto)

Writes proto to the file.

Args:
  proto:  A protocol buffer.

TFRecordWriter

A GenomicsWriter that writes to a TFRecord file.

Example usage:
  writer = TFRecordWriter('/tmp/my_output.tfrecord.gz')
  for record in records:
    writer.write(record)

Note that TFRecord files do not need to be wrapped in a "with" block.

Methods:

__init__(self, output_path, header=None, compression_type=None)

Initializer.

Args:
  output_path: str. The output path to which the records are written.
  header: An optional header for the particular data type. This can be
    useful for file types that have logical headers where some operations
    depend on that header information (e.g. VCF using its headers to
    determine type information of annotation fields).
  compression_type:  Either 'ZLIB', 'GZIP', '' (uncompressed), or
    None.  If None, __init__ will guess the compression type based on
    the input_path's suffix.

Raises:
  IOError:  if there was any problem opening output_path for writing.

close(self)

Explicitly closes writer.

write(self, proto)

Writes the proto to the TFRecord file.