neuralee 0.1.4

Scalable Visualization of Massive Single-Cell DataUsing Neural Networks

• Free software: MIT license

• Documentation: https://neuralee.readthedocs.io.

This is an applicable version for NeuralEE.

1. The datasets loading and preprocessing module is modified from scVI.

2. Define NeuralEE class and some auxiliary function, mainly for cuda computation, except like entropic affinity calculation which is quite faster computed on cpu.

3. General elastic embedding algorithm on cuda is given based on matlab code from Max Vladymyrov.

4. Add some demos of notebook helping to reproduce.

Installation

1. Install Python 3.7.

2. Install PyTorch. If you have an NVIDIA GPU, be sure to install a version of PyTorch that supports it. NeuralEE runs much faster with a discrete GPU.

3. Install NeuralEE through pip or from GitHub:

pip install neuralee

git clone git://github.com/HiBearME/NeuralEE.git
cd NeuralEE
python setup.py install --user

Tutorial

from neuralee.dataset import CortexDataset
from neuralee.embedding import NeuralEE

import torch

# detect whether to use GPU.
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')

# 1 load dataset.
cortex_dataset = CortexDataset(save_path='../data/')

# 2 preprocess dataset. logarithm transformation, genes subsample and standard scale.
cortex_dataset.log_shift()
cortex_dataset.subsample_genes(558)
cortex_dataset.standardscale()

# 3 embedding.
# 3.1 not using mini-batch trick, if dataset is not large.
# 3.1.1 calculate weights matrix
cortex_dataset.affinity()

# 3.1.2 initialize NeuralEE class.
NEE = NeuralEE(cortex_dataset, device=device)

# 3.1.3.1 elastic embedding.
results = NEE.EE()

# 3.1.3.2 NeuralEE.
results_Neural = NEE.fine_tune()

# 3.2 introduce mini-batch trick.
# 3.2.1 calculate weights matrix on each batch.
cortex_dataset.affinity_split(N_small=0.25)

# 3.2.2 initialize NeuralEE class.
NEE = NeuralEE(cortex_dataset, device=device)

# 3.2.3 elastic embedding.
results_Neural_with4batches = NEE.fine_tune()

For more detailed tutorials and reproduction of original paper’s results, check at notebook files.

Examples

HEMATO

BRAIN LARGE