Function metadata_to_proto

src/alphagenome/models/track_data_utils.py:96–190

Converts track metadata to a `TracksMetadata` protobuf message. Args: metadata: A pandas DataFrame containing track metadata, with the following required columns: name, strand. Returns: A `TracksMetadata` protobuf message.

(
    metadata: track_data.TrackMetadata,
)

def metadata_to_proto(
    metadata: track_data.TrackMetadata,
) -> dna_model_pb2.TracksMetadata:
  """Converts track metadata to a `TracksMetadata` protobuf message.

  Args:
    metadata: A pandas DataFrame containing track metadata, with the following
      required columns: name, strand.

  Returns:
    A `TracksMetadata` protobuf message.
  """
  names = metadata['name']
  default_values = [None] * len(names)

  columns = zip(
      metadata['name'],
      metadata['strand'],
      metadata.get('ontology_curie', default_values),
      metadata.get('biosample_type', default_values),
      metadata.get('biosample_name', default_values),
      metadata.get('biosample_life_stage', default_values),
      metadata.get('transcription_factor', default_values),
      metadata.get('histone_mark', default_values),
      metadata.get('gtex_tissue', default_values),
      metadata.get('Assay title', default_values),
      metadata.get('data_source', default_values),
      metadata.get('genetically_modified', default_values),
      metadata.get('endedness', default_values),
      metadata.get('nonzero_mean', default_values),
      strict=True,
  )

  metadata_protos = []
  for (
      name,
      strand,
      ontology_curie,
      biosample_type,
      biosample_name,
      biosample_life_stage,
      transcription_factor,
      histone_mark,
      gtex_tissue,
      assay,
      data_source,
      genetically_modified,
      endedness,
      nonzero_mean,
  ) in columns:
    if biosample_type:
      biosample = dna_model_pb2.Biosample(
          name=biosample_name,
          type=dna_model_pb2.BiosampleType.Value(
              f'BIOSAMPLE_TYPE_{biosample_type.upper()}'
          ),
          stage=biosample_life_stage,
      )