API reference

Hand-written reference for ProDEC's public classes. For a task-oriented walkthrough, see usage.md.

`ProteinDescriptors`

Loads and caches the catalog of available protein descriptors.

ProteinDescriptors(json: Optional[str] = None)

json — optional path to a custom descriptor JSON file. Defaults to the bundled data.json.

Member	Description
`available_descriptors` (property)	Sorted list of loaded descriptor IDs (or names, if a descriptor has no ID).
`update_available_descriptors(json)`	Reload descriptors from a path or file handle, replacing the currently loaded set.
`get_descriptor(id: str) -> Descriptor`	Instantiate (and cache) a `Descriptor` for the given ID.

Represents a single descriptor and computes raw per-residue values for a sequence.

Descriptor(desc_data: dict)

Normally instantiated via ProteinDescriptors.get_descriptor, not directly.

Member	Description
`summary` (property)	Dict of descriptor metadata: authors, year, journal, DOI, PMID, patent.
`definition` (property)	Tuple of `(scale names, scale values)` as defined for the descriptor.
`is_sequence_valid(sequence: str) -> bool`	Whether every residue in `sequence` is supported by this descriptor.
`get(sequence, flatten=True, gaps=0, prec=60, power=-4, dtype=np.float16, fast=False, **kwargs)`	Raw descriptor values for `sequence`. See `usage.md` for the meaning of each argument.
`pandas_get(sequences, ids=None, ..., nproc=None, quiet=False, ipynb=False)`	Same as `get`, computed over many sequences in parallel and returned as a `pandas.DataFrame`.

Enum identifying the transform to perform: AVG, ACC, PDT, FFT.

Member	Description
`fullname` (property)	Human-readable name of the transform.
`constant_length` (property)	Whether the transform produces output of constant length regardless of sequence length.
`binary` (property)	Whether the transform is restricted to binary descriptors.
`available()` (classmethod)	List of transform names and members that can be used to instantiate a `Transform`.

Applies a TransformType on top of a Descriptor's raw values.

Transform(type: Union[str, TransformType], descriptor: Descriptor)

Member	Description
`summary` (property)	Metadata dict for the current transform type.
`available_transforms()` (staticmethod)	Same as `TransformType.available()`.
`is_compatible(transform, descriptor) -> bool` (staticmethod)	Whether a transform type and a descriptor can be combined.
`get(sequence, flatten=True, lag=1, domains=2, normalize=True, **kwargs)`	Transformed values for `sequence`. `kwargs` are forwarded to `Descriptor.get`. See `usage.md`.
`pandas_get(sequences, ids=None, lag=1, domains=2, nproc=None, quiet=False, ipynb=False)`	Same as `get`, computed over many sequences in parallel and returned as a `pandas.DataFrame`.