API

Import Bering as:

import Bering as br

Objects

objects.Bering_Graph(df_spots_seg, ...[, ...])

Build Bering Object for training and prediction purposes.

Preprocess

preprocess.add_image_features(bg[, normalize])

Add image features to spots

Graph

graphs.BuildGraph(bg, df_spots[, ...])	Build nearest neighbor graph for mRNA/protein colocalization.
graphs.BuildGraph_fromRaw(bg, df_spots, ...)	Build Graphs from unsegmented spots
graphs.BuildWindowGraphs(bg[, ...])	Build Graphs for originally segemented cells.
graphs.CreateData(bg[, batch_size, ...])	Create training and testing data loader

Models

models.GCN(n_features, n_classes[, ...])	Node classification Model with Graph Convolutional Networks (GCN) and Multilayer Perceptron (MLP).
models.BaselineMLP(n_features, n_classes[, ...])	Node classification Model with Multilayer Perceptron (MLP) without using graph models.
models.GaussianSmearing([start, stop, ...])	This function is from torchdrug package.
models.ImageEncoder(image_dims[, ...])	Convolutional neural network to learn representation from staining images of different sizes.
models.EdgeClf(n_node_latent_features, image)	Edge classifier model which learns node classification embedding, image embedding and distance kernel

Training

training.TrainerNode([model, lr, ...])	Trainer for node classification model.
training.TrainerEdge(model, nodeclf_model, ...)	Trainer for edge classification model.
training.Training(bg[, ...])	Training both node classification and edge classification models.
training.EarlyStopper([patience, min_delta])	Early Stop implementation for loss values

Datasets

datasets.merfish_cortex_zhang([path])	Pre-processed MERFISH dataset from Zhang et al.
datasets.cosmx_nsclc_he([path])	Pre-processed Nanostring CoxMx dataset from He et al.
datasets.iss_ca1_qian([path])	Pre-processed ISS CA1 (sample 3-1 left) from Qian et al.
datasets.merfish_ileum_petukhov([path])	Pre-processed MERFISH dataset from Petukhov et al.
datasets.stereoseq_embryobrain_chen([path])	Pre-processed Stereo-seq dataset from Chen et al.
datasets.xenium_dcis_janesick([path])	Pre-processed Xenium dataset from Janesick et al.

Pretrained Models

datasets.model_merfish_cortex_zhang([path])	Pre-trained model of MERFISH Cortex dataset from Zhang et al.
datasets.model_cosmx_nsclc_he([path])	Pre-trained model of Nanostring ung cancer dataset from He et al.
datasets.model_iss_ca1_qian([path])	Pre-trained model of ISS CA1 hippocampus data (sample 3-1 left) from Qian et al.
datasets.model_xenium_dcis_janesick([path])	Pre-trained model of Xenium breast cancer dataset from Janesick et al.

Plotting

plotting.Plot_SliceImages(bg)	Plot the whole slice, with individual segmented cells as spots.
plotting.Plot_Classification(bg, cell_name)	Plot node classfication results on the original data and predicted data.
plotting.Plot_Segmentation(bg, cell_name[, ...])	Plot the segmentation results with cell IDs on the original data and predicted data.
plotting.Plot_Segmentation_Post(bg, location)	Plot Original Cell IDs and Cell ID distribution on latent space Either input a cell name and then extract a table or a pre-filtered spots window.
plotting.Plot_Classification_Post(bg, location)	Plot original spots and newly-segmented spots

Utilities

tools.node_classification(bg, df_spots[, ...])	Node classification for all spots in the slice
tools.cell_segmentation(bg[, use_image, ...])	Run cell segmentation for all spots.
tools.cell_annotation(bg[, min_transcripts, ...])	Annotate segmented cells based on ensemble strategy.
tools.cell_analyze(adata[, min_counts, ...])	Run a standard analysis pipeline from single-cell data with raw counts.