zjufanlab / scdeepsort Goto Github PK

I can't import this package, it seems the problem is with it's DGL dependency and potentially how it was compiled with Cython. I have reinstalled and I have also tried an upper version of DGL but it fails (and your requirements clearly say dgl==0.4.3)

I'm trying the cpu version

import deepsort

---

AttributeError Traceback (most recent call last)
/tmp/ipykernel_267/491137184.py in
1 #from deepsort import DeepSortPredictor
----> 2 import deepsort
3 #model = DeepSortPredictor(species='human', tissue='Liver')

~/.local/lib/python3.8/site-packages/deepsort/init.py in
----> 1 from .predict import DeepSortPredictor
2 from .trainer import DeepSortClassifier
3 from .demo import demo

~/.local/lib/python3.8/site-packages/deepsort/predict.py in
----> 1 import dgl
2 from dgl.contrib.sampling import NeighborSampler
3 import torch
4 import torch.nn as nn
5 import torch.nn.functional as F

~/.local/lib/python3.8/site-packages/dgl/init.py in
6 # Need to ensure that the backend framework is imported before load dgl libs,
7 # otherwise weird cuda problem happens
----> 8 from .backend import load_backend, backend_name
9
10 from . import function

~/.local/lib/python3.8/site-packages/dgl/backend/init.py in
72
73
---> 74 load_backend(get_preferred_backend())
75
76

~/.local/lib/python3.8/site-packages/dgl/backend/init.py in load_backend(mod_name)
21
22 def load_backend(mod_name):
---> 23 mod = importlib.import_module('.%s' % mod_name, name)
24 thismod = sys.modules[name]
25 for api in backend.dict.keys():

/usr/lib/python3.8/importlib/init.py in import_module(name, package)
125 break
126 level += 1
--> 127 return _bootstrap._gcd_import(name[level:], package, level)
128
129

~/.local/lib/python3.8/site-packages/dgl/backend/pytorch/init.py in
----> 1 from .tensor import *

~/.local/lib/python3.8/site-packages/dgl/backend/pytorch/tensor.py in
8 from torch.utils import dlpack
9
---> 10 from ... import ndarray as nd
11 from ... import kernel as K
12 from ...function.base import TargetCode

~/.local/lib/python3.8/site-packages/dgl/ndarray.py in
13
14 from ._ffi.object import register_object, ObjectBase
---> 15 from ._ffi.function import _init_api
16 from ._ffi.ndarray import DGLContext, DGLType, NDArrayBase
17 from ._ffi.ndarray import context, empty, from_dlpack, numpyasarray

~/.local/lib/python3.8/site-packages/dgl/_ffi/function.py in
14 raise ImportError()
15 if sys.version_info >= (3, 0):
---> 16 from ._cy3.core import _set_class_function, _set_class_module
17 from ._cy3.core import FunctionBase as _FunctionBase
18 from ._cy3.core import convert_to_dgl_func

dgl/_ffi/_cython/./object.pxi in init dgl._ffi._cy3.core()

AttributeError: type object 'dgl._ffi._cy3.core.ObjectBase' has no attribute 'reduce_cython'

Hi, Xin Shao!
Could you tell me how to save my own training model? So that, when I use the model next time, I don't need re-run the "fit" step （it takes too long😧), just running the "model.predict".
Thanks!!!

I have read the preview paper and found that was a pretty cool method to do classification.

But I have one questions, about the setting of learnable sharing parameter beta;
it that the fixed-parameter or need to calculated by another formula?
will it updated during the iteration?

Edit: I was providing the wrong input in test_files (I was giving labels instead of data)

some problem with the code, is that the version problem? Since i cannot install 1.4.0 verion of torch. only can install version 1.7.1. other pacakges version is the same.

(base) G:\gibh\SINGLE CELL\jiazi\predict_celltype\scDeepSort-updated_pretrained_models>python train.py --species human --tissue Adipose --gpu -1 --filetype gz
Using backend: pytorch
{'batch_size': 500,
'dense_dim': 400,
'dropout': 0.1,
'exclude_rate': 0.005,
'filetype': 'gz',
'gpu': -1,
'hidden_dim': 200,
'lr': 0.001,
'n_epochs': 300,
'n_layers': 1,
'num_neighbors': 0,
'random_seed': 10086,
'species': 'human',
'test_rate': 0.2,
'threshold': 0,
'tissue': 'Adipose',
'unsure_rate': 2.0,
'weight_decay': 0.0005}
The build graph contains 16397 genes with 4 labels supported.
human_Adipose3_data.gz -> Nonzero Ratio: 2.75%
Added 1354 nodes and 610321 edges.
#Nodes in Graph: 17751, #Edges: 1220642.
[PCA] Gene EVR: 66.25 %.
------Train label statistics------
#1 [Stromal cell]: 184
#2 [Fibroblast]: 53
#3 [Macrophage]: 1099
#4 [Monocyte]: 18
Train Number: 1084, Test Number: 270
Traceback (most recent call last):
File "train.py", line 169, in
trainer.fit()
File "train.py", line 47, in fit
loss = self.train()
File "train.py", line 71, in train
for batch, nf in enumerate(NeighborSampler(g=self.graph,
File "D:\anaconda\lib\site-packages\dgl\contrib\sampling\sampler.py", line 314, in init
super(NeighborSampler, self).init(
File "D:\anaconda\lib\site-packages\dgl\contrib\sampling\sampler.py", line 166, in init
self._seed_nodes = utils.toindex(self._seed_nodes)
File "D:\anaconda\lib\site-packages\dgl\utils.py", line 242, in toindex
return data if isinstance(data, Index) else Index(data)
File "D:\anaconda\lib\site-packages\dgl\utils.py", line 15, in init
self._initialize_data(data)
File "D:\anaconda\lib\site-packages\dgl\utils.py", line 22, in _initialize_data
self._dispatch(data)
File "D:\anaconda\lib\site-packages\dgl\utils.py", line 47, in _dispatch
raise DGLError('Index data must be an int64 vector, but got: %s' % str(data))
dgl._ffi.base.DGLError: Index data must be an int64 vector, but got: tensor([17413, 17532, 17478, ..., 16965, 17085, 16614], dtype=torch.int32)

hello, i want to know in train.py, seperate the data after PCA, or seperate the data into trainset and testest, then PCA individually?

I downloaded and installed scDeepSort-v1.0-cpu.tar.gz as described but encoutered an error that tensorflow 2.4.0 is not supported. I tried tensorflow 2.3.0 but the numpy version was not compatible though I tried kinds of versions from 1.16.0 to 1.21.0. Also, tensorflow 2.2.0 was not available with the error AttributeError: module 'tensorflow' has no attribute '__version__' .Could anyone help me with this issue?

The following error is reported; is there any hints to solve this

>>> deepsort.demo()
Error in py_call_impl(callable, dots$args, dots$keywords) : 
  AttributeError: module 'dgl.backend' has no attribute 'is_tensor'

Detailed traceback:
  File "/Users/zhangzhongheng/.virtualenvs/r-reticulate/lib/python3.8/site-packages/deepsort/demo.py", line 9, in demo
    result = predictor.predict(demo_file)
  File "/Users/zhangzhongheng/.virtualenvs/r-reticulate/lib/python3.8/site-packages/deepsort/predict.py", line 46, in predict
    test_dict = self.__load_test_data(input_files)
  File "/Users/zhangzhongheng/.virtualenvs/r-reticulate/lib/python3.8/site-packages/deepsort/predict.py", line 212, in __load_test_data
    test_graph_dict[test_file] = dgl.DGLGraph()
  File "/Users/zhangzhongheng/.virtualenvs/r-reticulate/lib/python3.8/site-packages/dgl/graph.py", line 1003, in __init__
    self._node_frame = FrameRef(Frame(num_rows=self.number_of_nodes()))
  File "/Users/zhangzhongheng/.virtualenvs/r-reticulate/lib/python3.8/site-packages/dgl/frame.py", line 499, in __init__
    self._index = utils.toindex(slice(0, self._frame.num_rows))
  F

Hi,

I am trying to train on my own data. But I get the error:

Segmentation fault (core dumped)

The build graph contains 43946 genes with 41 labels supported.
data.gz -> Nonzero Ratio: 19.50%
Added 17267 nodes and 148004641 edges.
#Nodes in Graph: 61213, #Edges: 296009282.
Segmentation fault (core dumped)

Any idea what could cause this error?

Best regards,
Mario

# packages in environment at /home/mario/miniconda3/envs/egervari:
#
# Name                    Version                   Build  Channel
_libgcc_mutex             0.1                 conda_forge    conda-forge
_openmp_mutex             4.5                       1_gnu    conda-forge
ca-certificates           2020.12.5            ha878542_0    conda-forge
certifi                   2020.12.5        py36h5fab9bb_0    conda-forge
cpuonly                   1.0                           0    pytorch
decorator                 4.4.2                      py_0    conda-forge
dgl                       0.4.3post2               py36_0    dglteam
freetype                  2.10.4               h7ca028e_0    conda-forge
intel-openmp              2020.2                      254  
joblib                    0.17.0                     py_0    conda-forge
jpeg                      9d                   h36c2ea0_0    conda-forge
lcms2                     2.11                 hcbb858e_1    conda-forge
ld_impl_linux-64          2.35.1               hed1e6ac_0    conda-forge
libblas                   3.9.0                3_openblas    conda-forge
libcblas                  3.9.0                3_openblas    conda-forge
libffi                    3.3                  h58526e2_2    conda-forge
libgcc-ng                 9.3.0               h5dbcf3e_17    conda-forge
libgfortran-ng            7.5.0               hae1eefd_17    conda-forge
libgfortran4              7.5.0               hae1eefd_17    conda-forge
libgomp                   9.3.0               h5dbcf3e_17    conda-forge
liblapack                 3.9.0                3_openblas    conda-forge
libopenblas               0.3.12          pthreads_hb3c22a3_1    conda-forge
libpng                    1.6.37               h21135ba_2    conda-forge
libstdcxx-ng              9.3.0               h2ae2ef3_17    conda-forge
libtiff                   4.1.0                h4f3a223_6    conda-forge
libwebp-base              1.1.0                h36c2ea0_3    conda-forge
lz4-c                     1.9.2                he1b5a44_3    conda-forge
mkl                       2020.2                      256  
ncurses                   6.2                  h58526e2_4    conda-forge
networkx                  2.5                        py_0    conda-forge
ninja                     1.10.2               h4bd325d_0    conda-forge
numpy                     1.17.2           py36h95a1406_0    conda-forge
olefile                   0.46               pyh9f0ad1d_1    conda-forge
openssl                   1.1.1h               h516909a_0    conda-forge
pandas                    0.25.1           py36hb3f55d8_0    conda-forge
pillow                    8.0.1            py36h10ecd5c_0    conda-forge
pip                       20.3.1             pyhd8ed1ab_0    conda-forge
python                    3.6.11          hffdb5ce_3_cpython    conda-forge
python-dateutil           2.8.1                      py_0    conda-forge
python_abi                3.6                     1_cp36m    conda-forge
pytorch                   1.4.0               py3.6_cpu_0  [cpuonly]  pytorch
pytz                      2020.4             pyhd8ed1ab_0    conda-forge
readline                  8.0                  he28a2e2_2    conda-forge
scikit-learn              0.22.2.post1     py36hcdab131_0    conda-forge
scipy                     1.3.1            py36h921218d_2    conda-forge
setuptools                49.6.0           py36h9880bd3_2    conda-forge
six                       1.15.0             pyh9f0ad1d_0    conda-forge
sqlite                    3.34.0               h74cdb3f_0    conda-forge
tk                        8.6.10               hed695b0_1    conda-forge
torchvision               0.5.0                  py36_cpu  [cpuonly]  pytorch
wheel                     0.36.1             pyhd3deb0d_0    conda-forge
xlrd                      1.2.0              pyh9f0ad1d_1    conda-forge
xz                        5.2.5                h516909a_1    conda-forge
zlib                      1.2.11            h516909a_1010    conda-forge
zstd                      1.4.5                h6597ccf_2    conda-forge

Hello,

I am trying to setup scDeepSort using CUDA 10.2.
I've downloaded scDeepSort-v1.0-cu102.tar.gz but when I install it via pip, I get the error:
Could not find a versionthat satisfies the requirement torch==1.6.0

I have tried python versions 3.9.7, 3.8.6, 3.8.0 and 3.7.0. All lead to the same error.
The lowest version I managed to install was 1.7.1 but this only installed the CPU version, even though I followed the official pytorch installation instructions (and CUDA 10.2 is correctly installed).

So my questions are the following:

• Is scDeepSort compatible with newer pytorch versions?
• Which python version is recommended to install torch==1.6.0?

Thank you and best regards,
Mario

Hi,
Doing repeated training on different training sets will delete previous trained models. Is there a workaround?
During training only about 25% of different celltypes are recognized as unique cell types no matter how many features are included (100 genes, 1000 genes,10000 genes, etc.). Is there a maxium of unique celltypes accepted during the training process?
Thanks a lot!

Hi，scDeepSort team,using GNN for cell type indentify is a pretty cool work. But after analyzing by the scDeepSort in my dataset, I had one confusions:
my data have 0.2M(200000) cells,scDeepSort's computation is very slow,may be scDeepSort need use all cells to build graph,After running ~12h and still not finishing, I finally gave up(CPU version).Therefore,I perform cell subtype identification for each cluster cells,and cell subtype identification by clusters cells alone greatly increases speed，But I doubt that this will lose the global cell subtype information.

I have installed the package and can import it, and I can run the code deepsort.demo(), but when running：
test_file = 'D:/rwork/allcount.csv'
predictor.predict(test_file, save_path='results')
it reports an error：
Traceback (most recent call last):
File "", line 1, in
NameError: name 'predictor' is not defined
Could anyone help me with this error？

Hello,

I noticed in the training process, the graph was constructed by adding edges (gene to cell) and (cell to gene) and self-loop.

However, in testing, I found the edges are original training edges ((gene to cell) and (cell to gene) and self-loop in training) plus (gene to test cell) and test cell self-loop. The test cell to gene edges ar missing. So I am wondering, is there a reason for this?

Or I misunderstand the code?

Thank you!

I noticed this tool very earlier and I very interested at it.

Hi, scDeepSort team!

I'm using scdeepsort-cpu to train my own model. But I encountered this problem when I run model.fit()

My cuda version is 11.5. The compatible dgl cuda version is >=0.9. But the scdeepsort-cpu requires dgl==0.4.3. I'm confused about the error of requirement for the cuda version of dgl, and I cannot install this version to solve the problem. Could you help me with this problem?

I got an error when importing deepsort. The existing solutions I've searched and found cannot solve this problem. What reason would it be?

Python 3.7.12 | packaged by conda-forge | (default, Oct 26 2021, 06:08:53)
[GCC 9.4.0] on linux
Type "help", "copyright", "credits" or "license" for more information.

import deepsort
Traceback (most recent call last):
File "", line 1, in
File "/home/yangyuan/miniconda3/envs/scDeepsort/lib/python3.7/site-packages/deepsort/init.py"line 1, in
from .predict import DeepSortPredictor
File "/home/yangyuan/miniconda3/envs/scDeepsort/lib/python3.7/site-packages/deepsort/predict.py",ine 1, in
import dgl
File "/home/yangyuan/miniconda3/envs/scDeepsort/lib/python3.7/site-packages/dgl/init.py", lin8, in
from .backend import load_backend, backend_name
File "/home/yangyuan/miniconda3/envs/scDeepsort/lib/python3.7/site-packages/dgl/backend/init.", line 74, in
load_backend(get_preferred_backend())
File "/home/yangyuan/miniconda3/envs/scDeepsort/lib/python3.7/site-packages/dgl/backend/init.", line 23, in load_backend
mod = importlib.import_module('.%s' % mod_name, name)
File "/home/yangyuan/miniconda3/envs/scDeepsort/lib/python3.7/importlib/init.py", line 127, iimport_module
return bootstrap.gcd_import(name[level:], package, level)
File "/home/yangyuan/miniconda3/envs/scDeepsort/lib/python3.7/site-packages/dgl/backend/tensorflo__init.py", line 4, in
from .tensor import *
File "/home/yangyuan/miniconda3/envs/scDeepsort/lib/python3.7/site-packages/dgl/backend/tensorflotensor.py", line 31, in
import tfdlpack
File "/home/yangyuan/miniconda3/envs/scDeepsort/lib/python3.7/site-packages/tfdlpack/init.py"line 2, in
from .core import version, to_dlpack, from_dlpack, get_device_and_dtype
File "/home/yangyuan/miniconda3/envs/scDeepsort/lib/python3.7/site-packages/tfdlpack/core.py", li 15, in
dlpack_ops = load_library.load_op_library(libinfo.find_lib_path(libname)[0])
File "/home/yangyuan/miniconda3/envs/scDeepsort/lib/python3.7/site-packages/tensorflow_core/pythoframework/load_library.py", line 61, in load_op_library
lib_handle = py_tf.TF_LoadLibrary(library_filename)
tensorflow.python.framework.errors_impl.NotFoundError: /home/yangyuan/miniconda3/envs/scDeepsort/lipython3.7/site-packages/tfdlpack/libtfdlpack-tf-2.1.0.so: undefined symbol: _ZN10tensorflow12OpDefBlder4AttrESs

Hello,

In the paper, aggregation process, it says there is a learnable sharing parameter beta. I didn't see beta in the code. Are you using another name for this parameter?

Thank you!

Could I retain the existed model with new data?
any function could provided?

Hi, Xin Shao!

I have three problems in running scDeepsort

First, if data has less than 400 cells, PCA error is occured. How can I solve it ?

Second, In my own prediction in human_pancreas_data, Cell_type data has alpha, beta, delta etc.
But original label is Alpha cell, Beta cell, Delta cell. So my own acc is 0.
In this case, Do I suucess prediction?

Last, Can I get a based reference or standard of mapping file?

Hello,
I tried to create my own model with some of your mouse data and was not able to run the test files with the predict-function.
I used the mouse Liver data.

from deepsort import DeepSortClassifier

define the model

model1 = DeepSortClassifier(species='mouse',
tissue='Liver',
dense_dim=50,
hidden_dim=20,
gpu_id=-1,
n_layers=2,
random_seed=1,
n_epochs=20)

fit the model
train_files = [(pathtrain +'/mouse_Liver261_data.csv',pathtrain +'/mouse_Liver261_celltype.csv'),
(pathtrain +'/mouse_Liver4424_data.csv', pathtrain +'/mouse_Liver4424_celltype.csv')]

model1.fit(train_files, save_path='/mnt/workspace/ckilian/scDStut/results')

until here everything works fine

use the saved model to predict
test_files = [pathtest + '/mouse_Liver18000_data.csv', pathtest + '/mouse_Liver18000_celltype.csv']

for test_file in test_files: model1.predict(test_file, save_path='../results', model_path='../results')

TypeError: '>' not supported between instances of 'str' and 'int'

I am not sure what exactly leads to the problem and I appreciate any help.

I can't overcome this assertion error:

~/.local/lib/python3.8/site-packages/deepsort/trainer.py in __load_train_data(self, files, save_dir)
    239             # filter out cells not in label-text
    240             df = df.iloc[filter_cell]
--> 241             assert cell2type['cell'].tolist() == df.index.tolist()
    242             df = df.rename(columns=gene2id)
    243             # filter out useless columns if exists (when using gene intersection)

AssertionError: 

This is probably because my input data does not perfectly match the format that it expects. How do the tutorial files ('/path/to/human_brain_data_1.csv', '/path/to/human_brain_celltype_1.csv') exactly look like?

Here are my inputs:

x_train.iloc[:,:3]

Hi, installed and trying to run the example provided but we are missing the map folder as it was not included in the git repo. Where can we find this?

Thanks!

The package scDeepSort-v1.0-cu102.tar.gz cannot be downloaded in the github page. It turns to the following page:

This XML file does not appear to have any style information associated with it. The document tree is shown below.
<Error>
<Code>SignatureDoesNotMatch</Code>
<Message>The request signature we calculated does not match the signature you provided. Check your key and signing method.</Message>
<AWSAccessKeyId>AKIAIWNJYAX4CSVEH53A</AWSAccessKeyId>
<StringToSign>AWS4-HMAC-SHA256 20210921T005934Z 20210921/us-east-1/s3/aws4_request 11854b7786adcd58d466715ec171a26028640572a0bf105997633d565e3eca3f</StringToSign>
<SignatureProvided>1e5043e9359ecd3e6e606abe2df8fef320c64cd72ae0c4630b7526dee9d0dc6b</SignatureProvided>
<StringToSignBytes>41 57 53 34 2d 48 4d 41 43 2d 53 48 41 32 35 36 0a 32 30 32 31 30 39 32 31 54 30 30 35 39 33 34 5a 0a 32 30 32 31 30 39 32 31 2f 75 73 2d 65 61 73 74 2d 31 2f 73 33 2f 61 77 73 34 5f 72 65 71 75 65 73 74 0a 31 31 38 35 34 62 37 37 38 36 61 64 63 64 35 38 64 34 36 36 37 31 35 65 63 31 37 31 61 32 36 30 32 38 36 34 30 35 37 32 61 30 62 66 31 30 35 39 39 37 36 33 33 64 35 36 35 65 33 65 63 61 33 66</StringToSignBytes>
<CanonicalRequest>GET /256732044/df7f6300-a1cc-11eb-8061-a3846d3775d8 X-Amz-Algorithm=AWS4-HMAC-SHA256&X-Amz-Credential=AKIAIWNJYAX4CSVEH53A%2F20210921%2Fus-east-1%2Fs3%2Faws4_request&X-Amz-Date=20210921T005934Z&X-Amz-Expires=300&X-Amz-SignedHeaders=host&actor_id=22423040&key_id=0&repo_id=256732044&response-content-disposition=attachment%3B%20filename%3DscDeepSort-v1.0-cu102.tar.gz&response-content-type=application%2Foctet-stream host:github-production-release-asset-2e65be.s3.amazonaws.com host UNSIGNED-PAYLOAD</CanonicalRequest>
<CanonicalRequestBytes>47 45 54 0a 2f 32 35 36 37 33 32 30 34 34 2f 64 66 37 66 36 33 30 30 2d 61 31 63 63 2d 31 31 65 62 2d 38 30 36 31 2d 61 33 38 34 36 64 33 37 37 35 64 38 0a 58 2d 41 6d 7a 2d 41 6c 67 6f 72 69 74 68 6d 3d 41 57 53 34 2d 48 4d 41 43 2d 53 48 41 32 35 36 26 58 2d 41 6d 7a 2d 43 72 65 64 65 6e 74 69 61 6c 3d 41 4b 49 41 49 57 4e 4a 59 41 58 34 43 53 56 45 48 35 33 41 25 32 46 32 30 32 31 30 39 32 31 25 32 46 75 73 2d 65 61 73 74 2d 31 25 32 46 73 33 25 32 46 61 77 73 34 5f 72 65 71 75 65 73 74 26 58 2d 41 6d 7a 2d 44 61 74 65 3d 32 30 32 31 30 39 32 31 54 30 30 35 39 33 34 5a 26 58 2d 41 6d 7a 2d 45 78 70 69 72 65 73 3d 33 30 30 26 58 2d 41 6d 7a 2d 53 69 67 6e 65 64 48 65 61 64 65 72 73 3d 68 6f 73 74 26 61 63 74 6f 72 5f 69 64 3d 32 32 34 32 33 30 34 30 26 6b 65 79 5f 69 64 3d 30 26 72 65 70 6f 5f 69 64 3d 32 35 36 37 33 32 30 34 34 26 72 65 73 70 6f 6e 73 65 2d 63 6f 6e 74 65 6e 74 2d 64 69 73 70 6f 73 69 74 69 6f 6e 3d 61 74 74 61 63 68 6d 65 6e 74 25 33 42 25 32 30 66 69 6c 65 6e 61 6d 65 25 33 44 73 63 44 65 65 70 53 6f 72 74 2d 76 31 2e 30 2d 63 75 31 30 32 2e 74 61 72 2e 67 7a 26 72 65 73 70 6f 6e 73 65 2d 63 6f 6e 74 65 6e 74 2d 74 79 70 65 3d 61 70 70 6c 69 63 61 74 69 6f 6e 25 32 46 6f 63 74 65 74 2d 73 74 72 65 61 6d 0a 68 6f 73 74 3a 67 69 74 68 75 62 2d 70 72 6f 64 75 63 74 69 6f 6e 2d 72 65 6c 65 61 73 65 2d 61 73 73 65 74 2d 32 65 36 35 62 65 2e 73 33 2e 61 6d 61 7a 6f 6e 61 77 73 2e 63 6f 6d 0a 0a 68 6f 73 74 0a 55 4e 53 49 47 4e 45 44 2d 50 41 59 4c 4f 41 44</CanonicalRequestBytes>
<RequestId>JRHAJ2QVZN8EKGR4</RequestId>
<HostId>3neTs5e7T8N48r7teCRl9aUlbi4Gmfo0aUtndtMnyQc7yekpnKtnaapItiyRTWwhxzC6HApIQhU=</HostId>
</Error>

But the scDeepSort-v1.0-cu100.tar.gz can be successfully downloaded.

How to train my own model?

I followed the preprocess script to created the .gz data (gene*cells ) and celltype table into csv (Celltype name is denifited by me)

then I put them into "mouse" file and the issue names is "project1"

But I get the information below:

The build graph contains 4944 genes with 0 labels supported. (there are 224 celltype labels)

I also find that there are not only 4 labels in file "human_Adipose1372_celltype" But why the information showed is :

The build graph contains 16397 genes with 4 labels supported.
human_Adipose3_data.gz -> Nonzero Ratio: 2.75%
Added 1354 nodes and 610321 edges.
#Nodes in Graph: 17751, #Edges: 1220642.
[PCA] Gene EVR: 66.25 %.
------Train label statistics------
#1 [Stromal cell]: 184
#2 [Macrophage]: 1099
#3 [Monocyte]: 18
#4 [Fibroblast]: 53

is the label information included into .gz file?