scHiCluster

https://zhoujt1994.github.io/scHiCluster/
Admin
scHiCluster is a comprehensive Python package for single-cell chromosome contact data analysis,...
Tokens:358,137
Snippets:2,435
Trust Score:9.7
Update:4 months ago
Show doc for...
Tokens:
Raw
### Command Example

Source: https://zhoujt1994.github.io/scHiCluster/_sources/command_line/hicluster_merge_cool.md

An example of how to use the hicluster merge-cool command.

```bash
hicluster merge-cool \
--input_cool_tsv_file impute/100K/cell_table_2.tsv \
--output_cool dataset/merge-cool.100kb
```

--------------------------------

### Command Line Example

Source: https://zhoujt1994.github.io/scHiCluster/_sources/command_line/hicluster_domain.md

An example of how to use the hicluster domain command with specified arguments.

```bash
hicluster domain \
--cell_table_path impute/25K/cell_table.tsv \
--output_prefix dataset/domain \
--resolution 25000 \
--window_size 10 \
--cpu 4 
```

--------------------------------

### Command Example

Source: https://zhoujt1994.github.io/scHiCluster/_sources/command_line/hicluster_contact_distance.md

An example of how to run the hicluster contact-distance command with specific parameters.

```bash
hicluster contact-distance \
--contact_table contact_table_rmbkl.tsv \
--chrom_size_path /data/aging/ref/m3C/mm10.main.nochrM.nochrY.chrom.sizes \
--output_prefix contact_distance \
--resolution 10000 \
--chr1 1 \
--pos1 2 \
--chr2 5 \
--pos2 6 \
--cpu 20

```

--------------------------------

### Command Example

Source: https://zhoujt1994.github.io/scHiCluster/_sources/command_line/hicluster_merge_cell_raw.md

This section provides an example of how to use the `hicluster merge-cell-raw` command with specific arguments.

```bash
hicluster merge-cell-raw \
--cell_table contact_table_rmbkl.tsv \
--chrom_size_path /data/aging/ref/m3C/mm10.main.nochrM.nochrY.chrom.sizes \
--output_file dataset/merge-cell-raw.5kb \
--resolution 5000 \
--chr1 1 \
--pos1 2 \
--chr2 5 \
--pos2 6 \
--min_pos_dist 2500
```

--------------------------------

### Loop Directory Setup

Source: https://zhoujt1994.github.io/scHiCluster/Tan2021/prepare.html

Sets up the output directory for loop calling.

```python
outdir = 'Tan2021_loop/'
loop_dir = f'/anvil/scratch/x-zhou/{outdir}'

```

--------------------------------

### hicluster cpg-ratio command example

Source: https://zhoujt1994.github.io/scHiCluster/_sources/command_line/hicluster_cpg_ratio.md

Example of calculating CpG ratio at each Chrom100K (resolution 100000).

```bash
hicluster cpg-ratio \
--fasta_path /data/aging/ref/m3C/mm10_with_chrl.fa \
--hdf_output_path cpg_ratio_100k.hdf \
--chrom_size_path /data/aging/ref/m3C/mm10.main.nochrM.nochrY.chrom.sizes \
--resolution 100000
```

--------------------------------

### Example 3: Imputation at 10kb resolution

Source: https://zhoujt1994.github.io/scHiCluster/_sources/command_line/hicluster_prepare_impute.md

This command prepares data for imputation at 10kb resolution, with adjusted window and step sizes compared to the previous examples.

```bash
hicluster prepare-impute \
    --cell_table contact_table_rmbkl.tsv \
    --batch_size 1536 \
    --pad 1 \
    --cpu_per_job 30 \
    --chr1 1 \
    --pos1 2 \
    --chr2 5 \
    --pos2 6 \
    --output_dir impute/100K/ \
    --chrom_size_path /data/aging/ref/m3C/mm10.main.nochrM.nochrY.chrom.sizes \
    --output_dist 5050000 \
    --window_size 30000000 \
    --step_size 10000000 \
    --resolution 10000 \

```

--------------------------------

### Example 2: Imputation at 25kb resolution

Source: https://zhoujt1994.github.io/scHiCluster/_sources/command_line/hicluster_prepare_impute.md

This command prepares data for imputation at 25kb resolution, similar to the 100kb example but with a different resolution setting.

```bash
hicluster prepare-impute \
    --cell_table contact_table_rmbkl.tsv \
    --batch_size 1536 \
    --pad 1 \
    --cpu_per_job 30 \
    --chr1 1 \
    --pos1 2 \
    --chr2 5 \
    --pos2 6 \
    --output_dir impute/100K/ \
    --chrom_size_path /data/aging/ref/m3C/mm10.main.nochrM.nochrY.chrom.sizes \
    --output_dist 5050000 \
    --window_size 500000000 \
    --step_size 500000000 \
    --resolution 25000 \

```

--------------------------------

### Example Usage

Source: https://zhoujt1994.github.io/scHiCluster/_sources/hba/domain_majortype/04.corr.ipynb

Illustrates how to use the Domain Major Type functionality.

```python
import numpy as np

def domain_majortype(data, n_components=None, n_neighbors=15, min_dist=0.1, metric='euclidean', random_state=None):
    """
    Perform Domain Major Type analysis on the input data.

    Parameters
    ----------
    data : array-like of shape (n_samples, n_features)
        The input data for analysis.
    n_components : int, optional
        The dimension of the embedded space.
    n_neighbors : int, optional
        The number of nearest neighbors to consider for manifold learning.
    min_dist : float, optional
        The minimum distance between points in the embedded space.
    metric : str, optional
        The distance metric to use for manifold learning.
    random_state : int or RandomState instance or None, optional
        Seed for the random number generator.

    Returns
    -------
    embedding : ndarray of shape (n_samples, n_components)
        The embedded representation of the data.
    graph : UMAP.graph.Graph
        The nearest neighbor graph.
    """
    if n_components is None:
        n_components = data.shape[1]

    # Perform UMAP for dimensionality reduction and visualization
    reducer = umap.UMAP(
        n_components=n_components,
        n_neighbors=n_neighbors,
        min_dist=min_dist,
        metric=metric,
        random_state=random_state,
    )
    embedding = reducer.fit_transform(data)
    graph = reducer.graph_

    return embedding, graph


def plot_domain_majortype(embedding, labels=None, title='Domain Major Type Analysis', ax=None):
    """
    Plot the results of Domain Major Type analysis.

    Parameters
    ----------
    embedding : ndarray of shape (n_samples, 2)
        The 2D embedding of the data.
    labels : array-like of shape (n_samples,), optional
        Labels for each data point.
    title : str, optional
        The title of the plot.
    ax : matplotlib.axes.Axes, optional
        The axes to plot on.
    """
    if ax is None:
        fig, ax = plt.subplots(figsize=(10, 8))

    scatter = ax.scatter(
        embedding[:, 0],
        embedding[:, 1],
        c=labels,
        cmap='viridis' if labels is None else 'tab20',
        s=10,
    )

    if labels is not None:
        legend = ax.legend(*scatter.legend_elements(), title='Domain Major Type')
        ax.add_artist(legend)

    ax.set_title(title)
    ax.set_xlabel('UMAP Dimension 1')
    ax.set_ylabel('UMAP Dimension 2')
    plt.show()


if __name__ == '__main__':
    # Example Usage:
    # Generate some synthetic data
    np.random.seed(42)
    data = np.random.rand(100, 10)
    labels = np.random.randint(0, 5, 100)

    # Perform Domain Major Type analysis
    embedding, graph = domain_majortype(data, n_components=2, random_state=42)

    # Plot the results
    plot_domain_majortype(embedding, labels=labels)

    # Example with no labels
    embedding_no_labels, _ = domain_majortype(data, n_components=2, random_state=42)
    plot_domain_majortype(embedding_no_labels, title='Domain Major Type Analysis (No Labels)')

```

--------------------------------

### Example Usage

Source: https://zhoujt1994.github.io/scHiCluster/_sources/hba/correlation_summary.ipynb

This snippet demonstrates how to generate a correlation summary.

```python
from schicluster import autoencoder
from schicluster import data
from schicluster import preprocess

# Load data
adata = data.load_anndata("path/to/your/data.h5ad")

# Preprocess data
adata = preprocess.preprocess_data(adata)

# Train autoencoder
autoencoder.train_autoencoder(adata)

# Generate correlation summary
autoencoder.correlation_summary(adata)
```

--------------------------------

### Output directory example

Source: https://zhoujt1994.github.io/scHiCluster/command_line/hicluster_prepare_impute.html

Specifies the output directory for the imputation preparation process.

```text
--output_dir impute/100K/

```

--------------------------------

### Example Usage

Source: https://zhoujt1994.github.io/scHiCluster/_sources/hba/compartment_majortype/05.corr.ipynb

Demonstrates how to use or interact with the compartment major type functionality.

```python
import os
import sys
import json
import numpy as np
import pandas as pd
from scipy.stats import pearsonr
from sklearn.cluster import KMeans
from sklearn.decomposition import PCA
from sklearn.preprocessing import StandardScaler
from sklearn.metrics import silhouette_score
from sklearn.ensemble import RandomForestClassifier
from sklearn.model_selection import train_test_split
from sklearn.metrics import accuracy_score, classification_report
from sklearn.manifold import TSNE
import matplotlib.pyplot as plt
import seaborn as sns

# Load data
data = pd.read_csv("../data/compartment_majorType.csv")
data.head()

# Data preprocessing
# Select features and target
X = data.iloc[:, 1:]
y = data.iloc[:, 0]

# Scale features
scaler = StandardScaler()
X_scaled = scaler.fit_transform(X)

# PCA for dimensionality reduction
pca = PCA(n_components=2)
X_pca = pca.fit_transform(X_scaled)

# K-Means clustering
k = 3  # Number of clusters
kmeans = KMeans(n_clusters=k, random_state=42, n_init=10)
kmeans.fit(X_pca)
clusters = kmeans.labels_

# Silhouette score for cluster evaluation
silhouette_avg = silhouette_score(X_pca, clusters)
print(f"Silhouette Score: {silhouette_avg}")

# Visualize clusters
plt.figure(figsize=(10, 7))
sns.scatterplot(x=X_pca[:, 0], y=X_pca[:, 1], hue=clusters, palette='viridis', s=100)
plt.title('K-Means Clustering of Compartment Major Types')
plt.xlabel('Principal Component 1')
plt.ylabel('Principal Component 2')
plt.legend(title='Cluster')
plt.grid(True)
plt.show()

# Train a classifier to predict major type
X_train, X_test, y_train, y_test = train_test_split(X_scaled, y, test_size=0.2, random_state=42)

model = RandomForestClassifier(n_estimators=100, random_state=42)
model.fit(X_train, y_train)

y_pred = model.predict(X_test)

print(f"Accuracy: {accuracy_score(y_test, y_pred)}")
print(classification_report(y_test, y_pred))

# t-SNE for visualization
tsne = TSNE(n_components=2, random_state=42, perplexity=30, n_iter=300)
X_tsne = tsne.fit_transform(X_scaled)

plt.figure(figsize=(10, 7))
sns.scatterplot(x=X_tsne[:, 0], y=X_tsne[:, 1], hue=y, palette='viridis', s=100)
plt.title('t-SNE Visualization of Compartment Major Types')
plt.xlabel('t-SNE Component 1')
plt.ylabel('t-SNE Component 2')
plt.legend(title='Major Type')
plt.grid(True)
plt.show()

```

--------------------------------

### Argument Parser Setup

Source: https://zhoujt1994.github.io/scHiCluster/_modules/schicluster/__main__.html

This snippet shows how arguments are added to the argument parser, including positional arguments and options for CPU usage.

```python
parser.add_argument('--pos1', type=int, default=2, required=False, 
                        help='0 based index of pos1 column.')
    parser.add_argument('--chr2', dest='chrom2', type=int, default=5, required=False, 
                        help='0 based index of chr2 column.')
    parser.add_argument('--pos2', type=int, default=6, required=False, 
                        help='0 based index of pos2 column.')
    parser.add_argument('--cpu', type=int, default=20, required=False, 
                        help='number of cpus to parallel.')
```

--------------------------------

### Region Selection

Source: https://zhoujt1994.github.io/scHiCluster/hba/plot_example.html

Defines the resolution and calculates the start and end loops for plotting.

```python
resl = 10000
loopl, loopr = (ll//resl), (rr//resl)
print(loopl, loopr)
```

--------------------------------

### Select Differential Compartment

Source: https://zhoujt1994.github.io/scHiCluster/hba/plot_example.html

Filters compartment data based on chromosome, start position, and statistical significance.

```python
resc = 100000
comp = pd.read_csv(f'{cdir}DifferentialResult/fdr_result/differential.intra_sample_combined.pcQnm.bedGraph', sep='	', header=0, index_col=None)
comp.index = comp['chr'] + '_' + (comp['start'] // resc).astype(str)
bincomp = comp[['chr', 'start', 'end', 'sample_maha', 'pval', 'padj']]
comp = comp[leg]
selc = (bincomp['chr']==chrom) & (bincomp['start']>=ll) & (bincomp['start']<rr) & (zscore(bincomp['sample_maha'])>norm.isf(0.025))
print(selc.sum())
```

--------------------------------

### Initial Setup and Directory Definitions

Source: https://zhoujt1994.github.io/scHiCluster/_sources/hba/loop_majortype/01.diff.ipynb

Sets up input and output directories and defines resolution and group name.

```python
indir = '/home/jzhou_salk_edu/sky_workdir/hba/loop_majortype/'
outdir = '/home/jzhou_salk_edu/sky_workdir/hba/loop_majortype/'
res = 10000
group = group_name
```

--------------------------------

### Select Differential Boundary

Source: https://zhoujt1994.github.io/scHiCluster/hba/plot_example.html

Filters boundaries based on chromosome, start position, and various quality/significance filters.

```python
binall = pd.read_hdf(f'{ddir}MajorType_boundary_stats.hdf', key='data')
selb = (binall['chrom']==chrom) & (binall['start']>=ll) & (binall['start']<rr) & binall['bklfilter'] & binall['chi2filter'] & binall['ins_lm'] & (binall['probdiff']>0.05) & binall['diff_sc']
print(selb.sum())
```

--------------------------------

### Figure and Axes Setup

Source: https://zhoujt1994.github.io/scHiCluster/hba/plot_example.html

This snippet sets up a matplotlib figure and axes with specific dimensions and grid specifications, designed to accommodate multiple subplots for plotting different data types. It also calculates initial positions for plotting elements.

```python
fig, axes = plt.subplots(len(leg)*2+1, 1, figsize=(10, np.sum([1] + np.tile([3.5,1],len(leg)).tolist())/2), 
                         gridspec_kw={'height_ratios':[1] + np.tile([3.5,1],len(leg)).tolist()}, dpi=300, sharex='all')

tmpl = loopall.loc[sell, [1,4]].values // resl - loopl
tmpd = binall.loc[selb, 'start'] // resl - loopl
tmpc = bincomp.loc[selc, 'start'] // resl - loopl

ax = axes[0]
sel = (binall['chrom']==chrom) & (binall['start']>=ll) & (binall['start']<rr)
xpos = (binall.loc[sel, 'start'] // resl - loopl) * np.sqrt(2)
ax.plot(xpos, binall.loc[sel, 'chi2_sc'].values, 
        c='C0', alpha=0.7)

ax.set_title('Chi-Square', fontsize=10)
ax.set_xlim([0, (loopr-loopl-1)*np.sqrt(2)])

for i in range(len(leg)):
    ax = axes[i*2+1]
    ax.set_title(legname[i], fontsize=10)
    ax.spines['right'].set_visible(False)
    ax.spines['top'].set_visible(False)
    ax.spines['bottom'].set_visible(False)
    ax.spines['left'].set_visible(False)
    img = ax.imshow(dstall[i], cmap='afmhot_r', vmin=0, vmax=0.012)
    h = len(dstall[i])
    ax.set_ylim([0.5*h, 0.4*h])
    ax.set_xlim([0, h])
    ax.set_yticks([])
    ax.set_yticklabels([])
    ax.scatter((tmpl[:, 0]+tmpl[:, 1])/np.sqrt(2), 0.5*h-(tmpl[:, 1]-tmpl[:, 0])/np.sqrt(2), 
           alpha=0.1, s=10, marker='D', edgecolors='c', color='none')

    ax = axes[i*2+2]
    ax.plot(xpos, bound_prob_ct.loc[leg[i], sel].values, c='C0', alpha=0.7)
    ax.scatter(tmpd * np.sqrt(2), np.zeros(len(tmpd))+0.05, color='r', s=4)
    ax.set_ylim([0, 0.3])
    ax2 = ax.twinx()
    ax2.plot(xpos, ins.loc[leg[i], sel].values, c='C1', alpha=0.7)
    ax2.set_ylim([0.05, 0.4])

ax.set_xlim([0, (loopr-loopl-1)*np.sqrt(2)])
ax.set_xticks(np.sqrt(2)*np.array(np.arange(0, loopr-loopl+1, 100).tolist() + [tss//resl-loopl, tes//resl-loopl]))
ax.set_xticklabels([])
ax.set_xticklabels([f'{(xx+loopl)/100}M' for xx in np.arange(0, loopr-loopl+1, 100)]+['TSS','TES'])
```

--------------------------------

### Principal Component Analysis initialization

Source: https://zhoujt1994.github.io/scHiCluster/hba/compartment_majortype/03.example.html

Initializes a PCA object from the scikit-learn library with a specified number of components.

```python
pca = PCA(n_components=2)
```

--------------------------------

### Install R Package

Source: https://zhoujt1994.github.io/scHiCluster/_modules/schicluster/domain/call_domain.html

Checks if an R package is installed and installs it if not.

```python
from rpy2 import robjects as r
from rpy2.robjects.packages import importr
from rpy2.robjects.vectors import StrVector


def install_r_package(name):
    if not isinstalled(name):
        utils = importr('utils')
        utils.chooseCRANmirror(ind=1)
        utils.install_packages(StrVector([name]))
```

--------------------------------

### Determine genomic region around a gene's TSS

Source: https://zhoujt1994.github.io/scHiCluster/hba/plot_example.html

Given a gene symbol ('LAMP5'), this code finds its chromosome, Transcription Start Site (TSS), and Transcription End Site (TES). It then defines a genomic window upstream (lslop) and downstream (rslop) of the TSS and prints the chromosome, window boundaries, and strand.

```python
gtmp = 'LAMP5'
lslop, rslop = 1000000, 2000000

chrom = genemeta.loc[gtmp, 0]
if genemeta.loc[gtmp, 5]=='+':
    tss = genemeta.loc[gtmp, 1]
    tes = genemeta.loc[gtmp, 2]
else:
    tss = genemeta.loc[gtmp, 2]
    tes = genemeta.loc[gtmp, 1]

ll, rr = (tss - lslop), (tss + rslop)
print(chrom, ll, rr, genemeta.loc[gtmp, 5])
```

--------------------------------

### Example contact_table.tsv format

Source: https://zhoujt1994.github.io/scHiCluster/command_line/hicluster_contact_distance.html

Example of the content for the contact_table.tsv file.

```text
cell_1 absolute_hic_rmbkl_contact_path_1
cell_2 absolute_hic_rmbkl_contact_path_2
cell_3 absolute_hic_rmbkl_contact_path_3

```

--------------------------------

### Define chromosome and binning parameters

Source: https://zhoujt1994.github.io/scHiCluster/hba/compartment_majortype/03.example.html

Sets the target chromosome and binning range, and determines the number of bins.

```python
chrom = 'chr5'
ll, rr = 800, 1200
n_bins = (int(chrom_sizes.loc[chrom]) // res + 1)
binfilter = binall[4]
```

--------------------------------

### Execute Snakemake for Step 1

Source: https://zhoujt1994.github.io/scHiCluster/_sources/Tan2021/prepare.ipynb

This command executes the Snakemake workflow generated for Step 1, utilizing 8 cores and continuing even if errors occur.

```bash
!snakemake --snakefile Snakefile -j 8 --keep-going
```

--------------------------------

### Data Loading and Preprocessing

Source: https://zhoujt1994.github.io/scHiCluster/hba/compartment_majortype/03.example.html

This snippet loads cooler files, extracts matrices, performs basic preprocessing (removing diagonal, filtering by bin), and applies PCA. It also correlates the first principal component with CpG density.

```python
Hall = []
Hpc = []
for i,ct in enumerate(['L5_IT', 'Sst', 'ASC', 'MGC']):
    cool = cooler.Cooler(f'{indir}{ct}/{ct}.CHcorr.cool')
    Q = cool.matrix(balance=False, sparse=True).fetch(chrom).toarray()
    Q = Q - np.diag(np.diag(Q))
    Q = Q[binfilter][:, binfilter]
    Hall.append(Q)
    pc = pca.fit_transform(Q)
    cpgtmp = cpg.loc[cpg['#1_usercol']==chrom, 'CpG_density'].values[binfilter]
    if pearsonr(cpgtmp, pc[:,0])[0]>0:
        pc = pc[:,0]
    else:
        pc = -pc[:,0]
    Hpc.append(pc)
    print(ct)
```

--------------------------------

### Directory and Chromosome Size Setup

Source: https://zhoujt1994.github.io/scHiCluster/_sources/hba/domain_majortype/03.diff.ipynb

Sets up input and output directories and reads chromosome sizes for analysis.

```python
indir = '/data/hba/domain_majortype/'
outdir = f'/home/jzhou_salk_edu/sky_workdir/hba/domain_majortype/diff/{group_name}/'

```

```python
chrom_size_path = '/home/jzhou_salk_edu/sky_workdir/hba/ref/hg38.main.chrom.sizes'
chrom_sizes = cooler.read_chromsizes(chrom_size_path, all_names=True)
chrom_sizes = chrom_sizes.iloc[:22]

```

--------------------------------

### Correlation Analysis Example

Source: https://zhoujt1994.github.io/scHiCluster/_sources/hba/compartment_majortype/05.corr.ipynb

Example of performing correlation analysis on compartment major types.

```python
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns

# Assuming 'df' is your DataFrame with compartment major type data
# Example DataFrame creation (replace with your actual data loading)
data = {
    'compartment_A': np.random.rand(100),
    'compartment_B': np.random.rand(100),
    'major_type_1': np.random.randint(0, 2, 100),
    'major_type_2': np.random.randint(0, 2, 100)
}
df = pd.DataFrame(data)

# Calculate the correlation matrix
correlation_matrix = df.corr()

# Plot the heatmap
plt.figure(figsize=(10, 8))
sns.heatmap(correlation_matrix, annot=True, cmap='coolwarm', fmt=".2f")
plt.title('Correlation Matrix of Compartment Major Types')
plt.show()

# Further analysis can be done on specific correlations
# For example, correlation between 'compartment_A' and 'major_type_1'
corr_A_MT1 = df['compartment_A'].corr(df['major_type_1'])
print(f"Correlation between compartment_A and major_type_1: {corr_A_MT1:.2f}")
```

--------------------------------

### Load and process CpG density data

Source: https://zhoujt1994.github.io/scHiCluster/hba/compartment_majortype/03.example.html

Loads CpG density data from a text file and calculates CpG density.

```python
cpg = pd.read_csv('/home/jzhou_salk_edu/sky_workdir/hba/ref/hg38.100kbin.CpG.txt', header=0, index_col=3, sep='\t')
cpg['CpG_density'] = cpg['14_user_patt_count'] / (cpg['13_seq_len'] - cpg['11_num_N'])
```

--------------------------------

### Chromosome Output Example

Source: https://zhoujt1994.github.io/scHiCluster/hba/compartment_majortype/01.call.html

Example output showing chromosome names and associated integer values.

```text
chr1 0
chr2 0
chr3 0
chr4 0
chr5 0
chr6 0
chr7 0
chr8 0
chr9 0
chr10 0
chr11 0
chr12 0
chr13 0
chr14 0
chr15 0
chr16 0
chr17 0
chr18 0
chr19 0
chr20 0
chr21 0
chr22 0
```

--------------------------------

### Prepare imputation with different resolutions

Source: https://zhoujt1994.github.io/scHiCluster/Tan2021/imputation.html

This snippet demonstrates how to prepare imputation by generating snakemake commands for different resolutions and batch sizes. It also shows how to aggregate these commands into a single file for submission.

```bash
# Generate a contact file list
ls rmbkl/* | awk -v p=$PWD '{printf("%s/%s\n", p, $0)}' > contact_table_rmbkl.txt
paste <(awk -F'/' '{print $NF}' contact_table_rmbkl.txt | cut -d. -f1) contact_table_rmbkl.txt | sort -k1,1 > contact_table_rmbkl.tsv

# Prepare imputation 
# Contact file list are divided into smaller chunks assigned to different jobs
# Commands to run imputation are generated
hicluster prepare-impute --cell_table contact_table_rmbkl.tsv --batch_size 1536 --pad 1 --cpu_per_job 96 --chr1 1 --pos1 2 --chr2 3 --pos2 4 --output_dir impute/100K/ --chrom_size_path /anvil/scratch/x-zhou/Tan2021/chrom_sizes.txt --output_dist 500000000 --window_size 500000000 --step_size 500000000 --resolution 100000
hicluster prepare-impute --cell_table contact_table_rmbkl.tsv --batch_size 1536 --pad 2 --cpu_per_job 96 --chr1 1 --pos1 2 --chr2 3 --pos2 4 --output_dir impute/25K/ --chrom_size_path /anvil/scratch/x-zhou/Tan2021/chrom_sizes.txt --output_dist 5050000 --window_size 500000000 --step_size 500000000 --resolution 25000
hicluster prepare-impute --cell_table contact_table_rmbkl.tsv --batch_size 1536 --pad 2 --cpu_per_job 96 --chr1 1 --pos1 2 --chr2 3 --pos2 4 --output_dir impute/10K/ --chrom_size_path /anvil/scratch/x-zhou/Tan2021/chrom_sizes.txt --output_dist 5050000 --window_size 30000000 --step_size 10000000 --resolution 10000
mkdir -p sbatch/impute/
cat impute/*/snakemake_cmd.txt > sbatch/impute/snakemake_cmd.txt

```

--------------------------------

### Command Example

Source: https://zhoujt1994.github.io/scHiCluster/_sources/command_line/hicluster_embedding.md

An example command to generate a chromosome-by-chromosome contact matrix using hicluster embedding.

```bash
hicluster embedding \
    --cell_table_path cell_table.tsv \
    --output_dir dataset/embedding \
    --dim 50 \
    --dist 1000000 \
    --resolution 100000 \
    --scale_factor 100000 \
    --norm_sig \
    --save_raw \
    --cpu 20 

```

--------------------------------

### Define input and output directories and resolution

Source: https://zhoujt1994.github.io/scHiCluster/hba/domain_majortype/02.count.html

Sets the paths for input and output data, and defines a resolution parameter.

```python
indir = '/data/hba/domain_majortype/'
outdir = '/home/jzhou_salk_edu/sky_workdir/hba/domain_majortype/'
res = 25000

```

--------------------------------

### Initialize Boundary Data

Source: https://zhoujt1994.github.io/scHiCluster/hba/domain_majortype/02.count.html

Initializes the boundary data with ones and calculates the boundary count based on non-zero elements in the raw data.

```python
sc_border.X.data = np.ones(sc_border.X.data.shape)
metadata.loc[sc_border.obs.index, 'boundary_count'] = sc_border.raw.X.getnnz(axis=1)
```

--------------------------------

### Command Example

Source: https://zhoujt1994.github.io/scHiCluster/_sources/command_line/hicluster_filter_contact.md

An example of processing snm3C-seq data of mouse brain using the hicluster filter-contact command.

```bash
hicluster filter-contact \
    --output_dir rmbkl \
    --blacklist_1d_path /data/aging/ref/m3C/mm10-blacklist.v2.bed.gz \
    --blacklist_2d_path /data/aging/ref/m3C/mm10_2d_blacklist.bedpe.gz \
    --cpu 20 \
    --chr1 1 \
    --pos1 2 \
    --chr2 5 \
    --pos2 6 \
    --contact_table contact_table.tsv \
    --chrom_size_path /data/aging/ref/m3C/mm10.main.nochrM.nochrY.chrom.sizes
```

--------------------------------

### Correlation Analysis Example

Source: https://zhoujt1994.github.io/scHiCluster/_sources/hba/compartment_majortype/05.corr.ipynb

Example R code for performing correlation analysis on compartment major types.

```r
library(tidyverse)
library(ggcorrplot)

# Load your data
data <- read.csv("your_data.csv")

# Select the columns for correlation analysis
correlation_data <- data %>% select(compartment_major_type_1, compartment_major_type_2, ...)

# Calculate the correlation matrix
cor_matrix <- cor(correlation_data, use = "complete.obs")

# Visualize the correlation matrix
ggcorrplot(cor_matrix, method = "circle", type = "lower", lab = TRUE, lab_size = 3, title = "Correlation Matrix of Compartment Major Types")

```

--------------------------------

### Example usage of hicluster gene-score

Source: https://zhoujt1994.github.io/scHiCluster/_sources/command_line/hicluster_gene_score.md

This is a practical example demonstrating how to use the hicluster gene-score command with specific file paths and parameters.

```bash
hicluster gene-score \
--cell_table_path impute/10K/cell_table.tsv \
--gene_meta_path /data/aging/ref/m3C/gencode.vM22.annotation.gene.sorted.bed.gz \
--resolution 10000 \
--output_hdf_path  geneimputescore.hdf \
--chrom_size_path /data/aging/ref/m3C/mm10.main.nochrM.nochrY.chrom.sizes \
--cpu 48 
--mode impute
```

--------------------------------

### Cell Table Path Example

Source: https://zhoujt1994.github.io/scHiCluster/_sources/command_line/hicluster_embedding.md

Example format for the cell_table.tsv file, specifying cell names and their corresponding absolute cool file paths.

```bash
cell_1 absolute_cool_file_path_1
cell_2 absolute_cool_file_path_2
cell_3 absolute_cool_file_path_3

```

--------------------------------

### Prepare directories and files for each cluster

Source: https://zhoujt1994.github.io/scHiCluster/_sources/Tan2021/prepare.ipynb

Iterates through each cluster, samples cells if necessary, and calls a function 'prepare_dir' to set up the directory structure and files for scHiCluster analysis.

```python
for cluster, sub_df in cell_table.groupby('cluster'):
    if sub_df.shape[0]>1500:
        tmp = sub_df.loc[np.random.choice(sub_df.index, 1500, False)]
    else:
        tmp = sub_df.copy()
    prepare_dir(f'{outdir}{cluster}/', tmp, GENERATE_MATRIX_CHUNK_TEMPLATE, params)
```

--------------------------------

### Contact Table Example

Source: https://zhoujt1994.github.io/scHiCluster/command_line/hicluster_filter_contact.html

An example of the contact table format, where the first column is the cell name and the second is the absolute path to the Hi-C contact file.

```text
cell_1  absolute_hic_contact_path_1
cell_2  absolute_hic_contact_path_2
cell_3  absolute_hic_contact_path_3
```

--------------------------------

### Define Parameters

Source: https://zhoujt1994.github.io/scHiCluster/_sources/Tan2021/prepare.ipynb

Defines parameters for resolution and chromosome size path.

```python
params = {
    'resolution': 100000,
    'chrom_size_path': '"/data/ref/mm10/genome/mm10.main20.chrom.sizes"',
}
```

--------------------------------

### Initialize variables for matrix creation

Source: https://zhoujt1994.github.io/scHiCluster/Tan2021/dipc_clustering.html

Records the start time and initializes an empty list for the matrix and sets the dimension for subsequent operations.

```python
start_time = time.time()
matrix = []
dim = 50
for chrom in chrom_sizes.index:
```
Tokens:
Raw