{
"cells": [
{
"cell_type": "code",
"execution_count": 1,
"metadata": {
"Collapsed": "false"
},
"outputs": [],
"source": [
"%%capture --no-stdout\n",
"import numpy as np\n",
"import matplotlib.pyplot as plt\n",
"import pandas as pd\n",
"import scanpy as sc\n",
"import cna\n",
"np.random.seed(0)"
]
},
{
"cell_type": "markdown",
"metadata": {
"Collapsed": "false"
},
"source": [
"# covarying neighborhood analysis tutorial"
]
},
{
"cell_type": "markdown",
"metadata": {
"Collapsed": "false"
},
"source": [
"This notebook demonstrates the basic use of the `cna` python package. Before you run it, make sure you either install `cna` by running `pip install cna` or by cloning the `cna` github repo and adding it to your python path. (In the latter case you'll likely need to manually install the `multianndata` package, which is a dependency of `cna`.)"
]
},
{
"cell_type": "markdown",
"metadata": {
"Collapsed": "false"
},
"source": [
"## table of contents:\n",
"1. [minimal working example](#first-section)\n",
"1. [processing data into the correct format](#second-section)\n",
"1. Additional CNA features: [additional results produced by cna's association function](#third-section)\n",
"1. Additional CNA features: [working directly with the NAM and NAM PCs](#fourth-section)"
]
},
{
"cell_type": "markdown",
"metadata": {
"Collapsed": "false"
},
"source": [
"## 1. minimal working example "
]
},
{
"cell_type": "markdown",
"metadata": {
"Collapsed": "false"
},
"source": [
"Let's run `cna` and visualize results for a toy dataset in the commonly used `AnnData` format that already has a UMAP nearest-neighbor graph. You can learn more about data pre-processing and formatting in Section 2."
]
},
{
"cell_type": "markdown",
"metadata": {
"Collapsed": "false"
},
"source": [
"### 1.1 read in data"
]
},
{
"cell_type": "markdown",
"metadata": {
"Collapsed": "false"
},
"source": [
"First, we'll read in the dataset and convert it to the format required for `cna`. This involves placing sample-level metadata in a dedicated pandas dataframe. (See Section 2 for more info)"
]
},
{
"cell_type": "code",
"execution_count": 2,
"metadata": {
"Collapsed": "false"
},
"outputs": [],
"source": [
"# read in data\n",
"# NB: if data are already in MultiAnnData format (see section 2), they can be read via `cna.read`\n",
"#. for example, try out: d = cna.read('data_multianndata.h5ad') instead of runnign this code block\n",
"\n",
"import anndata as ad\n",
"d = ad.read_h5ad('data_anndata.h5ad')\n",
"\n",
"from multianndata import MultiAnnData\n",
"d = MultiAnnData(d)\n",
"d.obs_to_sample(['case','male','batch'])"
]
},
{
"cell_type": "markdown",
"metadata": {
"Collapsed": "false"
},
"source": [
"### 1.2 perform case/control analysis"
]
},
{
"cell_type": "code",
"execution_count": 3,
"metadata": {
"Collapsed": "false"
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"qcd NAM not found; computing and saving\n",
"\ttaking step 1\n",
"\tmedian kurtosis: 16.40326205754668\n",
"\ttaking step 2\n",
"\tmedian kurtosis: 12.75799245750447\n",
"\ttaking step 3\n",
"\tmedian kurtosis: 6.1154849181504485\n",
"\ttaking step 4\n",
"\tmedian kurtosis: 3.177490071480447\n",
"stopping after 4 steps\n",
"throwing out neighborhoods with batch kurtosis >= 6\n",
"keeping 10000 neighborhoods\n",
"covariate-adjusted NAM not found; computing and saving\n",
"\twith ridge 100000.0 median batch kurtosis = 1.981293621817167\n",
"computing SVD\n",
"performing association test\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"/data/srlab1/yakir/py/cna/cna/tools/_association.py:74: UserWarning: global association p-value attained minimal possible value. Consider increasing Nnull\n",
" warnings.warn('global association p-value attained minimal possible value. '+\\\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"computing neighborhood-level FDRs\n",
"\n",
"global association p-value: 0.000999000999000999\n"
]
}
],
"source": [
"# perform association test for case/ctrl status, controlling for sex as a covariate and accounting for potential batch effect\n",
"res = cna.tl.association(d, #dataset\n",
" d.samplem.case, #sample-level attribute of intest (case/control status)\n",
" covs=d.samplem[['male']], #covariates to control for (in this case just one)\n",
" batches=d.samplem.batch) #batch assignments for each sample so that cna can account for batch effects\n",
"\n",
"print('\\nglobal association p-value:', res.p)"
]
},
{
"cell_type": "code",
"execution_count": 4,
"metadata": {
"Collapsed": "false"
},
"outputs": [
{
"data": {
"image/png": "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\n",
"text/plain": [
"
"
]
},
"metadata": {
"needs_background": "light"
},
"output_type": "display_data"
}
],
"source": [
"# visualize which populations are expanded or depleted among case samples relative to cntrls\n",
"cna.pl.umap_ncorr(d, #dataset\n",
" res, #cna results object\n",
" scatter0={'alpha':0.5, 's':20}, #plotting parameters for neighborhoods that pass FDR\n",
" scatter1={'alpha':0.05, 's':20}) #plotting parameters for neighborhoods that don't pass FDR\n",
"plt.title('p = {:.2e}'.format(res.p))\n",
"plt.show()"
]
},
{
"cell_type": "markdown",
"metadata": {
"Collapsed": "false"
},
"source": [
"In the plot above, each cell is colored according to its neighborhood coefficient, *i.e.*, color corresponds to the correlation of abundance in each cell's corresponding neighborhood to case/control status. Neighborhoods that don't pass FDR 5% are shown in gray. As this plot shows, cases have a greater abundance of cells from the left-hand cluster relative to controls."
]
},
{
"cell_type": "markdown",
"metadata": {
"Collapsed": "false"
},
"source": [
"### 1.3 analyze a more sophisticated phenotype"
]
},
{
"cell_type": "markdown",
"metadata": {
"Collapsed": "false"
},
"source": [
"Let's next perform an association test for male/female sex rather than case/control status..."
]
},
{
"cell_type": "code",
"execution_count": 5,
"metadata": {
"Collapsed": "false"
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"covariate-adjusted NAM not found; computing and saving\n",
"\twith ridge 100000.0 median batch kurtosis = 1.9949274142866247\n",
"computing SVD\n",
"performing association test\n",
"computing neighborhood-level FDRs\n"
]
},
{
"data": {
"image/png": "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\n",
"text/plain": [
"
"
]
},
"metadata": {
"needs_background": "light"
},
"output_type": "display_data"
}
],
"source": [
"# perform association test for male/female sex\n",
"# we control for case/control status as a covariate and account for potential batch effects\n",
"res = cna.tl.association(d, d.samplem.male, covs=d.samplem[['case']], batches=d.samplem.batch)\n",
"\n",
"# visualize the results\n",
"cna.pl.umap_ncorr(d, res,\n",
" scatter0={'alpha':0.5, 's':20},\n",
" scatter1={'alpha':0.05, 's':20})\n",
"plt.title('p = {:.2e}'.format(res.p))\n",
"plt.show()"
]
},
{
"cell_type": "markdown",
"metadata": {
"Collapsed": "false"
},
"source": [
"It's now clear that a sub-population of the left-hand cluster is more abundant in males. CNA automatically picked up on this without needing to be given a resolution parameter.\n",
"\n",
"Note that it looks like some cells in the right-hand cluster pass our FDR threshold of 5%. However, these cells comprise only ~6% of the cells that pass the threshold and so are likely false discoveries."
]
},
{
"cell_type": "markdown",
"metadata": {
"Collapsed": "false"
},
"source": [
"### 1.4 relate results to clusters"
]
},
{
"cell_type": "markdown",
"metadata": {
"Collapsed": "false"
},
"source": [
"If this dataset had been clustered, two clusters would likely have been defined."
]
},
{
"cell_type": "code",
"execution_count": 6,
"metadata": {
"Collapsed": "false"
},
"outputs": [
{
"data": {
"image/png": "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\n",
"text/plain": [
"
"
],
"text/plain": [
" corr_case\n",
"gene \n",
"b'gene_20' -0.100297\n",
"b'gene_21' -0.097420\n",
"b'gene_22' -0.089048\n",
"b'gene_14' -0.085773\n",
"b'gene_19' -0.079371"
]
},
"execution_count": 10,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"# view top 5 most negatively correlated genes\n",
"d.var.sort_values('corr_case', ascending=True).head()"
]
},
{
"cell_type": "markdown",
"metadata": {
"Collapsed": "false"
},
"source": [
"## 2. data formats and preprocessing "
]
},
{
"cell_type": "markdown",
"metadata": {
"Collapsed": "false"
},
"source": [
"Let's now understand better the data format used by `cna` and what pre-processing is necessary to run `cna`."
]
},
{
"cell_type": "markdown",
"metadata": {
"Collapsed": "false"
},
"source": [
"### 2.1 the `MultiAnnData` object"
]
},
{
"cell_type": "markdown",
"metadata": {
"Collapsed": "false"
},
"source": [
"The data format used by `cna` a python object called `MultiAnnData`, which is a derivative of the common `AnnData` object used by `scanpy` that is modified to explicitly accommodate multi-sample single-cell datasets by including sample-level metadata."
]
},
{
"cell_type": "markdown",
"metadata": {
"Collapsed": "false"
},
"source": [
"There are three ways to create `MultiAnnData` objects. Reading from `h5ad` format using `cna.read`, conversion from an `AnnData` object, and manual construction."
]
},
{
"cell_type": "markdown",
"metadata": {
"Collapsed": "false"
},
"source": [
"#### 2.1.1 read a `MultiAnnData` object from a file"
]
},
{
"cell_type": "code",
"execution_count": 11,
"metadata": {
"Collapsed": "false"
},
"outputs": [],
"source": [
"d = cna.read('data_multianndata.h5ad')"
]
},
{
"cell_type": "markdown",
"metadata": {
"Collapsed": "false"
},
"source": [
"Like an `AnnData` object, the `MultiAnnData` object `d` stores raw data in `.X`, per-cell data in `.obs`, and per-marker data in `.var`.\n",
"\n",
"However, `d` additional stores per-sample data in `d.samplem`. Let's take a look."
]
},
{
"cell_type": "code",
"execution_count": 12,
"metadata": {
"Collapsed": "false"
},
"outputs": [
{
"data": {
"text/html": [
"
\n",
"\n",
"
\n",
" \n",
"
\n",
"
\n",
"
case
\n",
"
male
\n",
"
batch
\n",
"
\n",
"
\n",
"
id
\n",
"
\n",
"
\n",
"
\n",
"
\n",
" \n",
" \n",
"
\n",
"
0
\n",
"
0
\n",
"
0
\n",
"
0
\n",
"
\n",
"
\n",
"
1
\n",
"
0
\n",
"
0
\n",
"
1
\n",
"
\n",
"
\n",
"
2
\n",
"
0
\n",
"
0
\n",
"
2
\n",
"
\n",
"
\n",
"
3
\n",
"
0
\n",
"
0
\n",
"
3
\n",
"
\n",
"
\n",
"
4
\n",
"
0
\n",
"
0
\n",
"
4
\n",
"
\n",
" \n",
"
\n",
"
"
],
"text/plain": [
" case male batch\n",
"id \n",
"0 0 0 0\n",
"1 0 0 1\n",
"2 0 0 2\n",
"3 0 0 3\n",
"4 0 0 4"
]
},
"execution_count": 12,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"d.samplem.head()"
]
},
{
"cell_type": "markdown",
"metadata": {
"Collapsed": "false"
},
"source": [
"#### 2.1.2 build a `MultiAnnData` object from an `AnnData` object"
]
},
{
"cell_type": "markdown",
"metadata": {
"Collapsed": "false"
},
"source": [
"The second way to create a `MultiAnnData` object is to create one from an `AnnData` object. In this case, you need to explicitly supply the sample-level metadata for `.samplem`. The easiest way to do this is typically to take this information from `.obs` if it is already stored there, as is often the case in single-cell datasets. Another option is to explicitly set `.samplem` to a `DataFrame` that's been manually created.\n",
"\n",
"**Either way, `cna` needs to know which column of `.obs` contains the sample id of each cell. `cna` assumes that this columns is named `id`, but this behavior can be changed by passing `sampleid='someothername'` into the `MultiAnnData` function. The name of this column also needs to match the name of the index of `.samplem`.**"
]
},
{
"cell_type": "code",
"execution_count": 13,
"metadata": {
"Collapsed": "false"
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"d.samplem has 0 columns\n",
"now d.samplem has 3 columns\n"
]
},
{
"data": {
"text/html": [
"
\n",
"\n",
"
\n",
" \n",
"
\n",
"
\n",
"
case
\n",
"
male
\n",
"
batch
\n",
"
\n",
"
\n",
"
id
\n",
"
\n",
"
\n",
"
\n",
"
\n",
" \n",
" \n",
"
\n",
"
0
\n",
"
0
\n",
"
0
\n",
"
0
\n",
"
\n",
"
\n",
"
1
\n",
"
0
\n",
"
0
\n",
"
1
\n",
"
\n",
"
\n",
"
2
\n",
"
0
\n",
"
0
\n",
"
2
\n",
"
\n",
"
\n",
"
3
\n",
"
0
\n",
"
0
\n",
"
3
\n",
"
\n",
"
\n",
"
4
\n",
"
0
\n",
"
0
\n",
"
4
\n",
"
\n",
" \n",
"
\n",
"
"
],
"text/plain": [
" case male batch\n",
"id \n",
"0 0 0 0\n",
"1 0 0 1\n",
"2 0 0 2\n",
"3 0 0 3\n",
"4 0 0 4"
]
},
"execution_count": 13,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"# read in AnnData object from h5ad file and convert it to MultiAnnData\n",
"d = ad.read_h5ad('data_anndata.h5ad')\n",
"d = MultiAnnData(d, sampleid='id') # the default value of sampleid is 'id', here we just repeat it for clarity\n",
"print('d.samplem has', len(d.samplem.columns), 'columns')\n",
"\n",
"# Since our metadata are stored per-cell, we need to copy the information that is actually per-sampl\n",
"# (case status, batch and covariates) to the per-sample metadata DataFrame `d.samplem`\n",
"d.obs_to_sample(['case','male','batch'])\n",
"print('now d.samplem has', len(d.samplem.columns), 'columns')\n",
"d.samplem.head()"
]
},
{
"cell_type": "markdown",
"metadata": {
"Collapsed": "false"
},
"source": [
"#### 2.1.3 build a `MultiAnnData` object manually"
]
},
{
"cell_type": "markdown",
"metadata": {
"Collapsed": "false"
},
"source": [
"Building a `MultiAnnData` object manually works just like building an `AnnData` object manually. The main difference is that you also need to tell `cna` which column of `.obs` contains the sample id of each cell. **`cna` will assume that this column is named `id` unless it is told otherwise.**"
]
},
{
"cell_type": "code",
"execution_count": 14,
"metadata": {
"Collapsed": "false"
},
"outputs": [],
"source": [
"import multianndata as mad\n",
"\n",
"# let's pretend we had X, obs, and var as standalone objects...\n",
"X, obs, var = d.X, d.obs, d.var\n",
"\n",
"# we can create a MultiAnnData object from these\n",
"d = mad.MultiAnnData(X=X, # expression matrix\n",
" obs=obs, # cell metadata matrix\n",
" sampleid='id') # name of column of cell_meta that contains sample id for each cell\n",
" # default value is 'id' but we repeat here for clarity\n",
"# convert per-cell metadata into per-sample metadata as before\n",
"d.obs_to_sample(['batch', 'case', 'male'])"
]
},
{
"cell_type": "markdown",
"metadata": {
"Collapsed": "false"
},
"source": [
"As above, if instead our raw metadata table already contained information per-sample, we could have created `d.samplem` directly."
]
},
{
"cell_type": "markdown",
"metadata": {
"Collapsed": "false"
},
"source": [
"### 2.2 pre-processing"
]
},
{
"cell_type": "markdown",
"metadata": {
"Collapsed": "false"
},
"source": [
"The main pre-processing step that is required before running `cna` is to compute a cell-cell similarity graph. Here we'll use `scanpy` to compute the standard UMAP similarity graph. (In principle, any graph can be computed here.)"
]
},
{
"cell_type": "code",
"execution_count": 15,
"metadata": {
"Collapsed": "false"
},
"outputs": [],
"source": [
"%%capture --no-stdout\n",
"\n",
"# compute the UMAP cell-cell similarity graph\n",
"sc.pp.neighbors(d)\n",
"\n",
"# compute UMAP coordinates for plotting\n",
"sc.tl.umap(d)\n",
"\n",
"# the following line would save the pre-processed data as a h5ad file\n",
"# d.write('test.h5ad')"
]
},
{
"cell_type": "markdown",
"metadata": {
"Collapsed": "false"
},
"source": [
"## 3. additional `cna` features"
]
},
{
"cell_type": "markdown",
"metadata": {
"Collapsed": "false"
},
"source": [
"### 3.1 additional results produced by the `cna` `association` function "
]
},
{
"cell_type": "markdown",
"metadata": {
"Collapsed": "false"
},
"source": [
"Let's re-do the `cna` analysis for case status and go over the results in more detail"
]
},
{
"cell_type": "code",
"execution_count": 16,
"metadata": {
"Collapsed": "false"
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"qcd NAM not found; computing and saving\n",
"\ttaking step 1\n",
"\tmedian kurtosis: 16.40326205754668\n",
"\ttaking step 2\n",
"\tmedian kurtosis: 12.75799245750447\n",
"\ttaking step 3\n",
"\tmedian kurtosis: 6.1154849181504485\n",
"\ttaking step 4\n",
"\tmedian kurtosis: 3.177490071480447\n",
"stopping after 4 steps\n",
"throwing out neighborhoods with batch kurtosis >= 6\n",
"keeping 10000 neighborhoods\n",
"covariate-adjusted NAM not found; computing and saving\n",
"\twith ridge 100000.0 median batch kurtosis = 1.981293621817167\n",
"computing SVD\n",
"performing association test\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"/data/srlab1/yakir/py/cna/cna/tools/_association.py:74: UserWarning: global association p-value attained minimal possible value. Consider increasing Nnull\n",
" warnings.warn('global association p-value attained minimal possible value. '+\\\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"computing neighborhood-level FDRs\n"
]
}
],
"source": [
"d = cna.read('data_multianndata.h5ad')\n",
"res = cna.tl.association(d, d.samplem.case, covs=d.samplem[['male']], batches=d.samplem.batch)"
]
},
{
"cell_type": "markdown",
"metadata": {
"Collapsed": "false"
},
"source": [
"The `res` object returned by the association function contains the following fields:\n",
"1. `p` is the global p-value for association\n",
"1. `k` is the number of NAM PCs used for the association test (automatically selected in a data-dependent way)\n",
"1. `ncorrs` is the vector of neighborhood coefficients\n",
"1. `fdr_10p_t` and `fdr_5p_t` contain the thresholds on `np.abs(ncorrs)` needed to provide FDRs of 10% and 5% respectively\n",
"1. `fdrs` contains information on neighborhood-level FDRs at other thresholds\n",
"1. `kept` is a boolean vector that specifies which cells passed qc. So, e.g., `ncorrs` is of length `kept.sum()`.\n",
"1. `r2` is the global prediction r-squared achieved by the model\n",
"1. `r2_perpc` is the prediction r-squared achieved by each pc on its own\n",
"1. `beta` is the coefficient vector, one coefficient per PC\n",
"1. `yresid` and `yresid_hat` are the phenotype and the predicted phenotype, both with covariates and batch residualized out"
]
},
{
"cell_type": "markdown",
"metadata": {
"Collapsed": "false"
},
"source": [
"The kurtosis numbers in the output above say how 'outlier-y' the neighborhood abundances are at each step in the diffusion that `cna` does to define its neighborhoods. The algorithm stops when these numbers stop going down quickly. But if you see that `cna` stops when kurtosis is still very high (like above 10), that's likely a sign that the data aren't qc'd properly or are otherwise unsuitable."
]
},
{
"cell_type": "markdown",
"metadata": {
"Collapsed": "false"
},
"source": [
"### 3.2 working directly with the NAM and NAM PCs "
]
},
{
"cell_type": "markdown",
"metadata": {
"Collapsed": "false"
},
"source": [
"The association function above also automatically computes the NAM and its SVD as a prerequisite to the association testing. These can be found in the following objects\n",
"1. `d.uns['NAM.T']` contains the transpose of the NAM (neighborhoods by samples)\n",
"1. `d.uns['NAM_sampleXpc']` contains the sample loadings of the principal components of the NAM\n",
"1. `d.uns['NAM_nbhdXpc']` contains the neighborhood loadings of the principal components of the NAM\n",
"1. `d.uns['NAM_svs']` contains the squared singular values of the NAM\n",
"\n",
"These objects can also be computed independent of any association test using the function `cna.tl.nam`, demonstrated below"
]
},
{
"cell_type": "code",
"execution_count": 17,
"metadata": {
"Collapsed": "false"
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"covariate-adjusted NAM not found; computing and saving\n",
"\twith ridge 100000.0 median batch kurtosis = 1.9949274142866247\n",
"computing SVD\n"
]
}
],
"source": [
"cna.tl.nam(d, batches=d.samplem.batch) #note: this NAM is being computed *without* adjusting for sex or case/ctrl status"
]
},
{
"cell_type": "markdown",
"metadata": {
"Collapsed": "false"
},
"source": [
"Let's visualize a few of the results..."
]
},
{
"cell_type": "markdown",
"metadata": {
"Collapsed": "false"
},
"source": [
"#### 3.2.1 neighborhood loadings of NAM PCs"
]
},
{
"cell_type": "markdown",
"metadata": {
"Collapsed": "false"
},
"source": [
"First, we can plot the neighborhood loadings of each NAM PC to see which abundance patterns they capture"
]
},
{
"cell_type": "code",
"execution_count": 18,
"metadata": {
"Collapsed": "false"
},
"outputs": [
{
"data": {
"image/png": "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\n",
"text/plain": [
"