API

Import popV as:

import popv

Preprocessing

For convenience we provide a class that processes query and reference dataset and creates a concatenated dataset. All relevant entries for annotation are stored in the uns object of the returned AnnData.

Import as:

from popv.preprocessing import Process_Query

preprocessing.Process_Query

Processes the query and reference dataset in preparation for the annotation pipeline.

Annotation pipeline

This is the core functionality of popV that performs annotation and consensus voting. AlgorithmsNT is a named tuple that summarizes all existing cell-type predictors.

Import as:

from popv.annotation import annotate_data

annotation.AlgorithmsNT	Dataclass to store all available algorithms.
annotation.annotate_data	Annotate an AnnData dataset preprocessed by popv.preprocessing.Process_Query by using the annotation pipeline.

Visualization

visualization.agreement_score_bar_plot	Create bar-plot of prediction scores in query cells after running popv.
visualization.prediction_score_bar_plot	Create bar-plot of prediction scores in query cells after running popv.
visualization.celltype_ratio_bar_plot	Create bar-plot of celltype rations in query as well as reference cells after running popv.
visualization.make_agreement_plots	Create plot of confusion matrix for different popv methods and consensus prediction.

Hub

Pretrained models are stored on HuggingFace and can be downloaded by using pull_from_huggingface_hub that returns a HubModel class.

hub.HubMetadata	Encapsulates the required metadata for popV hub models.
hub.HubModel	Wrapper for BaseModelClass backed by HuggingFace Hub.
hub.HubModelCardHelper	A helper for creating a ModelCard for popV hub models.
hub.create_criticism_report	Compute and store accuracy metrics for a model.

Algorithms

algorithms.BaseAlgorithm
algorithms.KNN_SCVI	Class to compute KNN classifier after scVI integration.
algorithms.SCANVI_POPV	Class to compute classifier in scANVI model.
algorithms.KNN_BBKNN	Class to compute KNN classifier after BBKNN integration.
algorithms.KNN_HARMONY	Class to compute KNN classifier after Harmony integration.
algorithms.Support_Vector	Class to compute LinearSVC.
algorithms.Random_Forest	Class to compute Random forest classifier.
algorithms.XGboost	Class to compute XGboost classifier.
algorithms.ONCLASS	Class to compute OnClass cell-type prediction.
algorithms.KNN_SCANORAMA	Class to compute KNN classifier after Scanorama integration.
algorithms.CELLTYPIST	Class to compute Celltypist classifier.