Transfer learning for spatial transcriptomics data and single-cell RNA-seq data.
Project description
stTransfer
Installation
pip install stTransfer
import stTransfer as st
st.dnn_workflow(data_path = '/data/input/single.h5ad',
ann_key = 'celltype', # celltype in adata.obs
marker_genes=None, # marker genes list
batch_size=4096, # train batch size
epochs=200, # train epochs
gpu="0", # gpu id
model_name="dnn.bgi", # model name
model_path="/data/model", # model path
filter_mt=False, # filter mitochondrial genes or not
cell_min_counts=300, # min counts per cell
gene_min_cells=10, # min cells per gene
cell_max_counts=98.) # max counts per cell
st_adata = st.load_data(data_path = '/data/input/st_adata.h5ad', # obsm.['spatial'] is required
filter_mt=True,
min_cells=10,
min_counts=300,
max_percent=98.0) # load data
st_adata_with_pslabel = st.transfer_from_sc_data(adata = st_adata, # adata with obsm.['spatial']
dnn_path = '/data/model/dnn.bgi', # dnn model path
gpu="0")
st.distribution_fine_tune(st_adata_with_pslabel,
pca_dim=0,
k_graph=30,
edge_weight=True,
epochs=100,
w_cls=50,
w_dae=1.,
w_gae=1.,
gpu="0",
save_path="/data/output") # output path
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
stTransfer-1.0.11.tar.gz
(11.0 kB
view hashes)
Built Distribution
Close
Hashes for stTransfer-1.0.11-py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | 579fd9c01200d870510b2554b7c6b6c2294cebc2f858b90a602a3f7b9dd7204b |
|
MD5 | 94a9217163aae1365ed85fc10692fc78 |
|
BLAKE2b-256 | 2f383b4389b223777c91b2064d263f7a8d4e5b8992fcd12f71d6b099bf2583a2 |