scvi.external.SCBASSET#
- class scvi.external.SCBASSET(adata, n_bottleneck_layer=32, l2_reg_cell_embedding=0.0, **model_kwargs)[source]#
Reimplementation of scBasset [Yuan and Kelley, 2022].
Performs representation learning of scATAC-seq data. Original implementation: calico/scBasset.
We are working to measure the performance of this model compared to the original.
- Parameters:
- adata
AnnData single-cell AnnData object that has been registered via
setup_anndata().- n_bottleneck_layer
int(default:32) Size of the bottleneck layer
- l2_reg_cell_embedding
float(default:0.0) L2 regularization for the cell embedding layer. A value, e.g. 1e-8 can be used to improve integration performance.
- **model_kwargs
Keyword args for
ScBassetModule
- adata
Examples
>>> adata = anndata.read_h5ad(path_to_sc_anndata) >>> scvi.data.add_dna_sequence(adata) >>> adata = adata.transpose() # regions by cells >>> scvi.external.SCBASSET.setup_anndata(adata, dna_code_key="dna_code") >>> model = scvi.external.SCBASSET(adata) >>> model.train() >>> adata.varm["X_scbasset"] = model.get_latent_representation()
Notes
See further usage examples in the following tutorials:
Attributes table#
Data attached to model instance. |
|
Manager instance associated with self.adata. |
|
The current device that the module's params are on. |
|
What the get normalized functions name is |
|
Returns computed metrics during training. |
|
Whether the model has been trained. |
|
Data attached to model instance. |
|
Returns the run id of the model. |
|
Returns the run name of the model. |
|
Summary string of the model. |
|
Observations that are in test set. |
|
Observations that are in train set. |
|
Observations that are in validation set. |
Methods table#
|
Converts a legacy saved model (<v0.15.0) to the updated save format. |
|
Returns the object in AnnData associated with the key in the data registry. |
|
Deregisters the |
|
Not implemented for this model class. |
|
Retrieves the |
Returns the cell-specific bias term. |
|
|
Returns the object in AnnData associated with the key in the data registry. |
Returns the latent representation of the cells. |
|
|
Not implemented for this model class. |
|
Returns the string provided to setup of a specific setup_arg. |
|
Returns the state registry for the AnnDataField registered with this instance. |
|
Infer transcription factor activity using a motif injection procedure. |
|
Variable names of input data. |
|
Instantiate a model from the saved output. |
|
Return the full registry saved with the model. |
|
Registers an |
|
Save the state of the model. |
|
Set up a model from on-disk AnnData files via the annbatch streaming loader. |
|
Sets up the |
|
Move the model to the device. |
|
Train the model. |
|
Transfer fields from a model to an AnnData object. |
|
Update setup method args. |
|
Print summary of the setup for the initial AnnData or a given AnnData object. |
|
Prints summary of the registry. |
|
Print args used to setup a saved model. |
Prints setup kwargs used to produce a given registry. |
Attributes#
- SCBASSET.DEFAULT_MOTIF_DIR = './scbasset_motifs/'#
- SCBASSET.MOTIF_URLS = {'human': ('https://storage.googleapis.com/scbasset_tutorial_data/Homo_sapiens_motif_fasta.tar.gz', 'Homo_sapiens_motif_fasta')}#
- SCBASSET.adata#
Data attached to model instance.
- SCBASSET.adata_manager#
Manager instance associated with self.adata.
- SCBASSET.device#
The current device that the module’s params are on.
- SCBASSET.get_normalized_function_name#
What the get normalized functions name is
- SCBASSET.history#
Returns computed metrics during training.
- SCBASSET.is_trained#
Whether the model has been trained.
- SCBASSET.registry#
Data attached to model instance.
- SCBASSET.run_id#
Returns the run id of the model. Used in MLFlow
- SCBASSET.run_name#
Returns the run name of the model. Used in MLFlow
- SCBASSET.summary_string#
Summary string of the model.
- SCBASSET.test_indices#
Observations that are in test set.
- SCBASSET.train_indices#
Observations that are in train set.
- SCBASSET.validation_indices#
Observations that are in validation set.
Methods#
- classmethod SCBASSET.convert_legacy_save(dir_path, output_dir_path, overwrite=False, prefix=None, **save_kwargs)[source]#
Converts a legacy saved model (<v0.15.0) to the updated save format.
- Parameters:
- dir_path
str Path to the directory where the legacy model is saved.
- output_dir_path
str Path to save converted save files.
- overwrite
bool(default:False) Overwrite existing data or not. If
Falseand directory already exists atoutput_dir_path, an error will be raised.- prefix
str|None(default:None) Prefix of saved file names.
- **save_kwargs
Keyword arguments passed into
save().
- dir_path
- Return type:
- SCBASSET.data_registry(registry_key)[source]#
Returns the object in AnnData associated with the key in the data registry.
- SCBASSET.deregister_manager(adata=None)[source]#
Deregisters the
AnnDataManagerinstance associated withadata.If
adataisNone, deregisters allAnnDataManagerinstances in both the class and instance-specific manager stores, except for the one associated with this model instance.
- SCBASSET.differential_abundance(*args, **kwargs)[source]#
Not implemented for this model class.
Available in models that inherit from
VAEMixin.- Raises:
- SCBASSET.get_anndata_manager(adata, required=False)[source]#
Retrieves the
AnnDataManagerfor a given AnnData object.Requires
self.idhas been set. Checks for anAnnDataManagerspecific to this model instance.- Parameters:
- Return type:
- SCBASSET.get_cell_bias()[source]#
Returns the cell-specific bias term.
- Return type:
- Returns:
bias (n_cells,)
- SCBASSET.get_from_registry(adata, registry_key)[source]#
Returns the object in AnnData associated with the key in the data registry.
AnnData object should be registered with the model prior to calling this function via the
self._validate_anndatamethod.
- SCBASSET.get_latent_representation()[source]#
Returns the latent representation of the cells.
- Return type:
- Returns:
latent representation (n_cells, n_latent)
- SCBASSET.get_normalized_expression(*args, **kwargs)[source]#
Not implemented for this model class.
Available in RNA models that inherit from
RNASeqMixin.- Raises:
- SCBASSET.get_setup_arg(setup_arg)[source]#
Returns the string provided to setup of a specific setup_arg.
- Return type:
- SCBASSET.get_state_registry(registry_key)[source]#
Returns the state registry for the AnnDataField registered with this instance.
- Return type:
- SCBASSET.get_tf_activity(tf, genome='human', motif_dir=None, lib_size_norm=True, batch_size=256)[source]#
Infer transcription factor activity using a motif injection procedure.
- Parameters:
- tf
str transcription factor name. must be provided in the relevant motif repository.
- genome
str(default:'human') species name for the motif injection procedure. Currently, only “human” is supported.
- motif_dir
str|None(default:None) path for the motif library. Will download if not already present.
- lib_size_norm
bool|None(default:True) normalized accessibility scores for library size by subtracting the cell bias term from each accessibility score prior to comparing motif scores to background scores.
- batch_size
int(default:256) minibatch size for TF activity inference.
- tf
- Return type:
- Returns:
tf_score [cells,] TF activity scores.
Notes
scBasset infers TF activities by injecting known TF motifs into a shuffled dinucleotide sequence and computing the change in accessibility predicted between the injected motif and a randomized background sequence. See [Yuan and Kelley, 2022] for details. We modeled this function off the original implementation in
scbasset.
- SCBASSET.get_var_names(legacy_mudata_format=False)[source]#
Variable names of input data.
- Return type:
- classmethod SCBASSET.load(dir_path, adata=None, accelerator='auto', device='auto', prefix=None, backup_url=None, datamodule=None, allowed_classes_names_list=None)[source]#
Instantiate a model from the saved output.
- Parameters:
- dir_path
str Path to saved outputs.
- adata
AnnData|MuData|None(default:None) AnnData organized in the same way as data used to train model. It is not necessary to run setup_anndata, as AnnData is validated against the saved
scvisetup dictionary. If None, will check for and load anndata saved with the model. If False, will load the model without AnnData.- accelerator
str(default:'auto') Supports passing different accelerator types
("cpu", "gpu", "tpu", "ipu", "hpu", "mps, "auto")as well as custom accelerator instances.- device
int|str(default:'auto') The device to use. Can be set to a non-negative index (
intorstr) or"auto"for automatic selection based on the chosen accelerator. If set to"auto"andacceleratoris not determined to be"cpu", thendevicewill be set to the first available device.- prefix
str|None(default:None) Prefix of saved file names.
- backup_url
str|None(default:None) URL to retrieve saved outputs from if not present on disk.
- datamodule
LightningDataModule|None(default:None) EXPERIMENTALALightningDataModuleinstance to use for training in place of the defaultDataSplitter. Can only be passed in if the model was not initialized withAnnData.- allowed_classes_names_list
list[str] |None(default:None) list of allowed classes names to be loaded (besides the original class name)
- dir_path
- Returns:
Model with loaded state dictionaries.
Examples
>>> model = ModelClass.load(save_path, adata) >>> model.get_....
- static SCBASSET.load_registry(dir_path, prefix=None)[source]#
Return the full registry saved with the model.
- classmethod SCBASSET.register_manager(adata_manager)[source]#
Registers an
AnnDataManagerinstance with this model class.Stores the
AnnDataManagerreference in a class-specific manager store. Intended for use in thesetup_anndata()class method followed up by retrieval of theAnnDataManagervia the_get_most_recent_anndata_manager()method in the model init method.Notes
Subsequent calls to this method with an
AnnDataManagerinstance referring to the same underlying AnnData object will overwrite the reference to previousAnnDataManager.
- SCBASSET.save(dir_path, prefix=None, overwrite=False, save_anndata=False, save_kwargs=None, legacy_mudata_format=False, datamodule=None, **anndata_write_kwargs)[source]#
Save the state of the model.
Neither the trainer optimizer state nor the trainer history are saved. Model files are not expected to be reproducibly saved and loaded across versions until we reach version 1.0.
- Parameters:
- dir_path
str Path to a directory.
- prefix
str|None(default:None) Prefix to prepend to saved file names.
- overwrite
bool(default:False) Overwrite existing data or not. If
Falseand directory already exists atdir_path, an error will be raised.- save_anndata
bool(default:False) If True, also saves the anndata
- save_kwargs
dict|None(default:None) Keyword arguments passed into
save().- legacy_mudata_format
bool(default:False) If
True, saves the modelvar_namesin the legacy format if the model was trained with aMuDataobject. The legacy format is a flat array with variable names across all modalities concatenated, while the new format is a dictionary with keys corresponding to the modality names and values corresponding to the variable names for each modality.- datamodule
LightningDataModule|None(default:None) EXPERIMENTALALightningDataModuleinstance to use for training in place of the defaultDataSplitter. Can only be passed in if the model was not initialized withAnnData.- anndata_write_kwargs
Kwargs for
write()
- dir_path
- classmethod SCBASSET.setup_annbatch(collection_path=None, paths=None, batch_key=None, labels_key=None, sample_key=None, unlabeled_category='Unknown', layer=None, categorical_covariate_keys=None, continuous_covariate_keys=None, rebuild=True, batch_size=4096, chunk_size=256, preload_nchunks=32, preload_to_gpu=True, dataset_size='20GB', shuffle=False, var_subset=None, merge=None, adatas=None, use_class_sampler=False, class_sampler_key=None, class_weights=None)[source]#
Set up a model from on-disk AnnData files via the annbatch streaming loader.
Builds (or reuses) a zarr-backed
DatasetCollectionfrom the supplied h5ad file paths, then wraps it in aAnnbatchDataModuleready for training.- Parameters:
- collection_path
str|None(default:None) Directory where the zarr collection is written (or already exists). If
None(default), a path is auto-generated as"./{ModelName}_annbatch.zarr"in the current working directory.- paths
list[str] |None(default:None) Paths to h5ad files that make up the training dataset. If
None, an existing collection atcollection_pathis opened without rebuilding — useful when the zarr store was created by a previous call. Must be provided when the store does not exist yet.- batch_key
str|None(default:None) Column in
obsto use as the batch variable.- labels_key
str|None(default:None) Column in
obsto use as the cell-type / label variable.- sample_key
str|None(default:None) Column in
obsto use as the sample variable. Used by models like MrVI.- unlabeled_category
str(default:'Unknown') Value used to mark unlabeled cells in
labels_key. Required by semi-supervised models such as SCANVI. Defaults to"Unknown".- layer
str|None(default:None) Layer in the h5ad files to use as the count matrix.
Noneusesadata.X(falling back toadata.raw.Xwhen a raw slot exists).- categorical_covariate_keys
list[str] |None(default:None) Additional categorical covariate columns in
obs.- continuous_covariate_keys
list[str] |None(default:None) Additional continuous covariate columns in
obs.- rebuild
bool(default:True) If
True, always rebuild the zarr collection even if it already exists on disk. IfFalse(default), reuse an existing collection and skip the (potentially expensive)add_adatasstep.- batch_size
int(default:4096) Number of cells per batch yielded by the
Loader.- chunk_size
int(default:256) Number of cells loaded from disk contiguously per read.
- preload_nchunks
int(default:32) Number of chunks to preload and shuffle in memory.
- preload_to_gpu
bool(default:True) Whether the loader should move data to GPU before yielding.
- dataset_size
int|str(default:'20GB') Number of observations to load into memory for shuffling / pre-processing when building the collection, or annbatch’s human-readable size strings, e.g.
"20GB".- shuffle
bool(default:False) Whether to pre-shuffle cells when building the collection.
- var_subset
list[str] |None(default:None) Optional list of gene names to restrict the collection to (passed as
var_subsettoadd_adatas).- merge
Literal['same','unique','first','only'] |None(default:None) How annbatch should merge
varmetadata across inputs. Passed through toannbatch.DatasetCollection.add_adatas().
- collection_path
- Returns:
AnnbatchDataModuleA configured datamodule whoseregistrycan be passed directly to the model constructor, e.g.model = SCVI(registry=dm.registry).
Notes
After training, saving and reloading the model requires reconstructing the datamodule manually and passing it to
load()via thedatamoduleargument — the same pattern used by all custom datamodules.
- classmethod SCBASSET.setup_anndata(adata, dna_code_key, layer=None, batch_key=None, **kwargs)[source]#
Sets up the
AnnDataobject for this model.A mapping will be created between data fields used by this model to their respective locations in adata. None of the data in adata are modified. Only adds fields to adata.
- Parameters:
- adata
AnnData AnnData object. Rows represent cells, columns represent features.
- dna_code_key
str Key in
adata.obsmwith dna sequences encoded as integer code.- layer
str|None(default:None) if not
None, uses this as the key inadata.layersfor raw count data.- batch_key
str|None(default:None) key in
adata.varfor batch information. Categories will automatically be converted into integer categories and saved toadata.var['_scvi_batch']. IfNone, assigns the same batch to all the data.
- adata
Notes
The adata object should be in the regions by cells’ format. This is due to scBasset considering regions as observations and cells as variables. This can be simply achieved by transposing the data,
bdata = adata.transpose().
- SCBASSET.to_device(device)[source]#
Move the model to the device.
- Parameters:
Examples
>>> adata = scvi.data.synthetic_iid() >>> model = scvi.model.SCVI(adata) >>> model.to_device("cpu") # moves model to CPU >>> model.to_device("cuda:0") # moves model to GPU 0 >>> model.to_device(0) # also moves model to GPU 0
- SCBASSET.train(max_epochs=1000, lr=0.01, accelerator='auto', devices='auto', train_size=None, validation_size=None, shuffle_set_split=True, batch_size=128, early_stopping=True, early_stopping_monitor='auroc_train', early_stopping_mode='max', early_stopping_min_delta=1e-06, datasplitter_kwargs=None, plan_kwargs=None, **trainer_kwargs)[source]#
Train the model.
- Parameters:
- max_epochs
int(default:1000) Number of epochs to train for
- lr
float(default:0.01) Learning rate for optimization.
- accelerator
str(default:'auto') Supports passing different accelerator types
("cpu", "gpu", "tpu", "ipu", "hpu", "mps, "auto")as well as custom accelerator instances.- devices
int|list[int] |str(default:'auto') The devices to use. Can be set to a non-negative index (
intorstr), a sequence of device indices (listor comma-separatedstr), the value-1to indicate all available devices, or"auto"for automatic selection based on the chosenaccelerator. If set to"auto"andacceleratoris not determined to be"cpu", thendeviceswill be set to the first available device.- train_size
float|None(default:None) Size of the training set in the range [0.0, 1.0].
- validation_size
float|None(default:None) Size of the test set. If
None, defaults to 1 -train_size. Iftrain_size + validation_size < 1, the remaining cells belong to a test set.- shuffle_set_split
bool(default:True) Whether to shuffle indices before splitting. If
False, the val, train, and test set are split in the sequential order of the data according tovalidation_sizeandtrain_sizepercentages.- batch_size
int(default:128) Minibatch size to use during training.
- early_stopping
bool(default:True) Perform early stopping. Additional arguments can be passed in
**kwargs. SeeTrainerfor further options.- early_stopping_monitor
str(default:'auroc_train') Metric logged during validation set epoch. The available metrics will depend on the training plan class used. We list the most common options here in the typing.
- early_stopping_mode
Literal['min','max'] (default:'max') In ‘min’ mode, training will stop when the quantity monitored has stopped decreasing, and in ‘max’ mode it will stop when the quantity monitored has stopped increasing.
- early_stopping_min_delta
float(default:1e-06) Minimum change in the monitored quantity to qualify as an improvement, i.e., an absolute change of less than min_delta, will count as no improvement.
- datasplitter_kwargs
dict|None(default:None) Additional keyword arguments passed into
DataSplitter.- plan_kwargs
dict|None(default:None) Keyword args for
TrainingPlan. Keyword arguments passed totrain()will overwrite values present inplan_kwargs, when appropriate.- **trainer_kwargs
Other keyword args for
Trainer.
- max_epochs
- SCBASSET.transfer_fields(adata, **kwargs)[source]#
Transfer fields from a model to an AnnData object.
- Return type:
- SCBASSET.update_setup_method_args(setup_method_args)[source]#
Update setup method args.
- Parameters:
- setup_method_args
dict This is a bit of a misnomer, this is a dict representing kwargs of the setup method that will be used to update the existing values in the registry of this instance.
- setup_method_args
- SCBASSET.view_anndata_setup(adata=None, hide_state_registries=False)[source]#
Print summary of the setup for the initial AnnData or a given AnnData object.