{
"cells": [
{
"cell_type": "markdown",
"id": "d6495895",
"metadata": {},
"source": [
"# Getting started (for datasets with unaligned features)\n",
"\n",
"Import the required packages:"
]
},
{
"cell_type": "code",
"execution_count": 1,
"id": "6935113b",
"metadata": {},
"outputs": [],
"source": [
"import os\n",
"import sys\n",
"from pathlib import Path\n",
"\n",
"import numpy as np\n",
"import pandas as pd\n",
"from matplotlib import pyplot as plt # optional\n",
"import seaborn as sns # optional\n",
"\n",
"import scanpy as sc\n",
"from scipy import sparse\n",
"\n",
"import networkx as nx\n",
"import torch"
]
},
{
"cell_type": "markdown",
"id": "79dade22",
"metadata": {},
"source": [
"If you get trouble with installing CAME, you can download the source code from [GitHub](https://github.com/XingyanLiu/CAME.git), \n",
"and append the path to `sys.path`. \n",
"\n",
"For example:\n",
"\n",
"```python\n",
"CAME_ROOT = Path('path/to/CAME')\n",
"sys.path.append(str(CAME_ROOT))\n",
"```"
]
},
{
"cell_type": "code",
"execution_count": 2,
"id": "c20e3fe2",
"metadata": {},
"outputs": [
{
"name": "stderr",
"output_type": "stream",
"text": [
"Using backend: pytorch\n"
]
}
],
"source": [
"import came\n",
"from came import pipeline, pp, pl\n",
"\n",
"ROOT = Path(\".\") # set root"
]
},
{
"cell_type": "markdown",
"id": "7396c5e5",
"metadata": {},
"source": [
"## 0 Load datasets\n",
"\n",
"### 0.1 Load the example datasets"
]
},
{
"cell_type": "code",
"execution_count": 3,
"id": "00d1ffd3",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"dict_keys(['adatas', 'varmap', 'varmap_1v1', 'dataset_names', 'key_class'])\n",
"a new directory made:\n",
"\t_temp\\('Baron_human', 'Baron_mouse')-(12-16 18.36.34)\\figs\n"
]
}
],
"source": [
"from came import load_example_data\n",
"\n",
"example_data_dict = load_example_data()\n",
"print(example_data_dict.keys())\n",
"\n",
"adatas = example_data_dict['adatas']\n",
"dsnames = example_data_dict['dataset_names']\n",
"df_varmap = example_data_dict['varmap']\n",
"df_varmap_1v1 = example_data_dict['varmap_1v1']\n",
"\n",
"adata_raw1, adata_raw2 = adatas\n",
"key_class1 = key_class2 = example_data_dict['key_class']\n",
"\n",
"# setting directory for results\n",
"time_tag = came.make_nowtime_tag()\n",
"resdir = ROOT /'_temp' / f'{dsnames}-{time_tag}'\n",
"figdir = resdir / 'figs'\n",
"came.check_dirs(figdir) # check and make the directory"
]
},
{
"cell_type": "markdown",
"id": "b134821e",
"metadata": {},
"source": [
"### 0.2 Load your own datasets\n",
"\n",
"To load your own datasets, see the code example below:\n",
"\n",
"```python\n",
"# ========= customized paths ==========\n",
"\n",
"dsnames = ('Baron_human', 'Baron_mouse') # the dataset names, set by user\n",
"dsn1, dsn2 = dsnames\n",
"\n",
"path_rawdata1 = CAME_ROOT / 'came/sample_data/raw-Baron_human.h5ad'\n",
"path_rawdata2 = CAME_ROOT / 'came/sample_data/raw-Baron_mouse.h5ad'\n",
"```\n",
"Path to homologous gene mappings, a dataframe with 2-columns, where \n",
"the first column stores the reference genes and the second \n",
"column stores the query ones.\n",
"\n",
"Note that `path_varmap_1v1` is optional (can be `None`). If not provided, the 1-to-1 homologous mappings will be inferred from the whole mappings.\n",
"\n",
"```python\n",
"path_varmap = CAME_ROOT / f'came/sample_data/gene_matches_human2mouse.csv'\n",
"path_varmap_1v1 = CAME_ROOT / f'came/sample_data/gene_matches_1v1_human2mouse.csv'\n",
"\n",
"```\n",
"\n",
"Load scRNA-seq datasets.\n",
"\n",
"```python\n",
"# ========= load data =========\n",
"df_varmap = pd.read_csv(path_varmap)\n",
"df_varmap_1v1 = pd.read_csv(path_varmap_1v1) if path_varmap_1v1 else came.pp.take_1v1_matches(df_varmap)\n",
"\n",
"adata_raw1 = sc.read_h5ad(path_rawdata1)\n",
"adata_raw2 = sc.read_h5ad(path_rawdata2)\n",
"adatas = [adata_raw1, adata_raw2]\n",
"```\n",
"\n",
"Sepcifiy the column names of the cell-type labels, where `key_class1` is for reference data, and `key_class2` is for query data. If there aren't any cell-type or clustering labels for the query cells, you can set `key_class=None`.\n",
"\n",
"```python\n",
"key_class1 = 'cell_ontology_class' # set by user\n",
"key_class2 = 'cell_ontology_class' # set by user\n",
"```\n",
"\n",
"Setting directory for results\n",
"\n",
"```python\n",
"time_tag = came.make_nowtime_tag()\n",
"resdir = ROOT /'_temp' / f'{dsnames}-{time_tag}' # set by user\n",
"figdir = resdir / 'figs'\n",
"came.check_dirs(figdir) # check and make the directory\n",
"```"
]
},
{
"cell_type": "markdown",
"id": "14ef61ed",
"metadata": {},
"source": [
"Filtering genes (a preprocessing step, optional)\n",
"\n",
"```python\n",
"sc.pp.filter_genes(adata_raw1, min_cells=3)\n",
"sc.pp.filter_genes(adata_raw2, min_cells=3)\n",
"```"
]
},
{
"cell_type": "markdown",
"id": "6b9aa256",
"metadata": {},
"source": [
"### 0.3 Inspect the compositions of different classes"
]
},
{
"cell_type": "code",
"execution_count": 4,
"id": "9242f5e1",
"metadata": {},
"outputs": [
{
"data": {
"text/html": [
"\n",
"\n",
"
\n",
" \n",
" \n",
" | \n",
" Baron_human | \n",
" Baron_mouse | \n",
"
\n",
" \n",
" \n",
" \n",
" | B cell | \n",
" NaN | \n",
" 10.0 | \n",
"
\n",
" \n",
" | Schwann cell | \n",
" 13.0 | \n",
" 6.0 | \n",
"
\n",
" \n",
" | T cell | \n",
" 7.0 | \n",
" 7.0 | \n",
"
\n",
" \n",
" | endothelial cell | \n",
" 252.0 | \n",
" 139.0 | \n",
"
\n",
" \n",
" | leukocyte | \n",
" NaN | \n",
" 8.0 | \n",
"
\n",
" \n",
" | macrophage | \n",
" 55.0 | \n",
" 36.0 | \n",
"
\n",
" \n",
" | mast cell | \n",
" 25.0 | \n",
" NaN | \n",
"
\n",
" \n",
" | pancreatic A cell | \n",
" 2326.0 | \n",
" 191.0 | \n",
"
\n",
" \n",
" | pancreatic D cell | \n",
" 601.0 | \n",
" 218.0 | \n",
"
\n",
" \n",
" | pancreatic PP cell | \n",
" 255.0 | \n",
" 41.0 | \n",
"
\n",
" \n",
" | pancreatic acinar cell | \n",
" 958.0 | \n",
" NaN | \n",
"
\n",
" \n",
" | pancreatic ductal cell | \n",
" 1077.0 | \n",
" 275.0 | \n",
"
\n",
" \n",
" | pancreatic epsilon cell | \n",
" 18.0 | \n",
" NaN | \n",
"
\n",
" \n",
" | pancreatic stellate cell | \n",
" 457.0 | \n",
" 61.0 | \n",
"
\n",
" \n",
" | type B pancreatic cell | \n",
" 2525.0 | \n",
" 894.0 | \n",
"
\n",
" \n",
"
\n",
"