diff --git a/envs/scarches_mapping_conda_env.yml b/envs/scarches_mapping_conda_env.yml index 0189047..5e5e238 100644 --- a/envs/scarches_mapping_conda_env.yml +++ b/envs/scarches_mapping_conda_env.yml @@ -7,10 +7,10 @@ dependencies: - pip - pip: - jupyterlab - - scanpy==1.8.2 - - torch==1.8.0 + - scanpy>=1.8.2 + - torch>=1.3,<=1.8.0 - scarches==0.3.5 - scvi-tools==0.8.1 - - umap-learn==0.5.2 - - pynndescent==0.5.5 + - umap-learn>=0.5.2 + - pynndescent #>=0.5.5 diff --git a/notebooks/LCA_scArches_mapping_new_data_to_hlca.ipynb b/notebooks/LCA_scArches_mapping_new_data_to_hlca.ipynb index 1e5c4a0..7437c05 100644 --- a/notebooks/LCA_scArches_mapping_new_data_to_hlca.ipynb +++ b/notebooks/LCA_scArches_mapping_new_data_to_hlca.ipynb @@ -11,14 +11,14 @@ "cell_type": "markdown", "metadata": {}, "source": [ - "In this notebook, we will guide you through how to map your data to the Human Lung Cell Atlas (core reference), perform label transfer, and more. For that purpose we use scArches, a method to map new single cell/single nucleus data to an existing reference (see also Lotfollahi et al., Nature Biotechnology 2021 https://doi.org/10.1038/s41587-021-01001-7). " + "In this notebook, we will guide you through how to map your data to the [Human Lung Cell Atlas](https://www.biorxiv.org/content/10.1101/2022.03.10.483747v1) (core reference), perform label transfer, and more. For that purpose we use scArches, a method to map new single cell/single nucleus data to an existing reference (see also [Lotfollahi et al., Nature Biotechnology 2021](https://doi.org/10.1038/s41587-021-01001-7)). " ] }, { "cell_type": "markdown", "metadata": {}, "source": [ - "Import the needed modules. __Note that we use scArches version 0.3.5.__ For efficiency of knn-graph and umap calculation, we recommend using scanpy>=1.8.2, umap-learn>0.5, and installing pynndescent: `pip install pynndescent`." + "Import the needed modules. __Note that we use scArches version 0.3.5. and scvi-tools 0.8.1__ For efficiency of knn-graph and umap calculation, we recommend using scanpy>=1.8.2, umap-learn>0.5, and installing pynndescent: `pip install pynndescent`. If you used the conda environment provided on our GitHub repo, all of these packages were automatically installed with the correct versions, so no need to check!" ] }, { @@ -211,7 +211,7 @@ "pydevd_tracing NA\n", "pygments 2.11.2\n", "pyparsing 3.0.7\n", - "pytz 2021.3\n", + "pytz 2022.1\n", "requests 2.27.1\n", "rich NA\n", "scarches NA\n", @@ -246,7 +246,7 @@ "Linux-3.10.0-1160.42.2.el7.x86_64-x86_64-with-centos-7.9.2009-Core\n", "112 logical CPU cores, x86_64\n", "-----\n", - "Session information updated at 2022-03-18 16:46\n", + "Session information updated at 2022-03-21 11:43\n", "\n" ] } @@ -326,8 +326,8 @@ "name": "stderr", "output_type": "stream", "text": [ - "2022-03-18 16:46:57 URL:https://zenodo.org/record/6337966/files/HLCA_emb_and_metadata.h5ad [217785664/217785664] -> \"HLCA_emb_and_metadata.h5ad\" [1]\n", - "2022-03-18 16:46:59 URL:https://zenodo.org/record/6337966/files/HLCA_reference_model.zip [5321666/5321666] -> \"HLCA_reference_model.zip\" [1]\n" + "2022-03-21 11:44:31 URL:https://zenodo.org/record/6337966/files/HLCA_emb_and_metadata.h5ad [217785664/217785664] -> \"HLCA_emb_and_metadata.h5ad\" [1]\n", + "2022-03-21 11:44:35 URL:https://zenodo.org/record/6337966/files/HLCA_reference_model.zip [5321666/5321666] -> \"HLCA_reference_model.zip\" [1]\n" ] } ], @@ -459,7 +459,7 @@ { "data": { "text/plain": [ - "('../test/testmeta.csv.gz', )" + "('../test/testmeta.csv.gz', )" ] }, "execution_count": 15, @@ -810,15 +810,15 @@ "\u001b[34mINFO \u001b[0m Training Unsupervised Trainer for \u001b[1;36m400\u001b[0m epochs. \n", "\u001b[34mINFO \u001b[0m Training SemiSupervised Trainer for \u001b[1;36m500\u001b[0m epochs. \n", "\u001b[34mINFO \u001b[0m KL warmup for \u001b[1;36m400\u001b[0m epochs \n", - "Training...: 57%|█████▋ | 283/500 [02:19<02:00, 1.80it/s]\u001b[34mINFO \u001b[0m Reducing LR on epoch \u001b[1;36m283\u001b[0m. \n", - "Training...: 57%|█████▋ | 285/500 [02:21<02:01, 1.77it/s]\u001b[34mINFO \u001b[0m \n", + "Training...: 57%|█████▋ | 283/500 [02:04<01:34, 2.30it/s]\u001b[34mINFO \u001b[0m Reducing LR on epoch \u001b[1;36m283\u001b[0m. \n", + "Training...: 57%|█████▋ | 285/500 [02:05<01:34, 2.28it/s]\u001b[34mINFO \u001b[0m \n", " Stopping early: no improvement of more than \u001b[1;36m0.001\u001b[0m nats in \u001b[1;36m10\u001b[0m epochs \n", "\u001b[34mINFO \u001b[0m If the early stopping criterion is too strong, please instantiate it with different \n", " parameters in the train method. \n", - "Training...: 57%|█████▋ | 285/500 [02:21<01:46, 2.01it/s]\n", + "Training...: 57%|█████▋ | 285/500 [02:05<01:35, 2.26it/s]\n", "\u001b[34mINFO \u001b[0m Training is still in warming up phase. If your applications rely on the posterior \n", " quality, consider training for more epochs or reducing the kl warmup. \n", - "\u001b[34mINFO \u001b[0m Training time: \u001b[1;36m52\u001b[0m s. \u001b[35m/\u001b[0m \u001b[1;36m500\u001b[0m epochs \n" + "\u001b[34mINFO \u001b[0m Training time: \u001b[1;36m46\u001b[0m s. \u001b[35m/\u001b[0m \u001b[1;36m500\u001b[0m epochs \n" ] } ],