### Identify Differentially Methylated Cell-Types with CellDMC

Source: https://context7.com/sjczheng/epidish/llms.txt

This example shows how to use the `CellDMC` function to identify differentially methylated cell-types (DMCTs). It requires beta values, a phenotype vector, and estimated cell-type fractions. P-value adjustment methods and thresholds can be specified.

```r
library(EpiDISH)

# Load data and estimate cell fractions first
data(centEpiFibIC.m)
data(DummyBeta.m)

# Step 1: Estimate cell-type fractions
out <- epidish(DummyBeta.m, centEpiFibIC.m, method = "RPC")
frac.m <- out$estF

# Step 2: Define phenotype vector (case/control)
pheno.v <- rep(c(0, 1), each = 5)  # 5 controls, 5 cases

# Step 3: Run CellDMC analysis
celldmc_result <- CellDMC(
  beta.m = DummyBeta.m,            # Beta values (CpGs x samples)
  pheno.v = pheno.v,               # Phenotype vector (binary or continuous)
  frac.m = frac.m,                 # Cell-type fractions from epidish
  adjPMethod = "fdr",              # P-value adjustment method
  adjPThresh = 0.05,               # Threshold for calling DMCTs
  cov.mod = NULL,                  # Optional covariates design matrix
  sort = FALSE,                    # Sort results by p-value
  mc.cores = 1                     # Number of cores for parallelization
)

# Access DMCT matrix: 0 = not DMC/DMCT, 1 = hyper, -1 = hypo
dmct_matrix <- celldmc_result$dmct
print(head(dmct_matrix))
```

```r
# Access detailed coefficients for each cell-type
epi_results <- celldmc_result$coe$Epi
print(head(epi_results))
```

```r
# With covariates adjustment (e.g., age and gender)
pheno_data <- data.frame(
  age = c(45, 52, 38, 61, 55, 48, 59, 42, 67, 51),
  gender = factor(c(rep("M", 5), rep("F", 5)))
)
cov_matrix <- model.matrix(~ age + gender, data = pheno_data)

celldmc_adjusted <- CellDMC(
  beta.m = DummyBeta.m,
  pheno.v = pheno.v,
  frac.m = frac.m,
  adjPMethod = "BH",
  adjPThresh = 0.05,
  cov.mod = cov_matrix,            # Adjust for age and gender
  sort = TRUE,                     # Sort by p-value
  mc.cores = 4                     # Use 4 cores
)

# Find significant DMCTs for each cell-type
sig_epi <- which(celldmc_adjusted$dmct[, "Epi"] != 0)
sig_fib <- which(celldmc_adjusted$dmct[, "Fib"] != 0)
sig_ic <- which(celldmc_adjusted$dmct[, "IC"] != 0)
```

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

### Load Reference and Beta Data in R

Source: https://github.com/sjczheng/epidish/blob/devel/README.md

Load the required reference centroid matrix and dummy beta value matrix for analysis.

```R
data(centEpiFibIC.m)
data(DummyBeta.m)
```

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

### Load EpiDISH library and data

Source: https://github.com/sjczheng/epidish/blob/devel/README.md

Loads the EpiDISH package and necessary data objects for cell-type fraction estimation.

```R
library(EpiDISH)
data(centDHSbloodDMC.m)
data(LiuDataSub.m)
```

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

### Display Session Information

Source: https://github.com/sjczheng/epidish/blob/devel/README.md

Output the current R session environment details.

```R
sessionInfo()
```

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

### Run CBS Method with hepidish

Source: https://context7.com/sjczheng/epidish/llms.txt

This snippet demonstrates how to use the `hepidish` function with the 'CBS' method to estimate cell-type fractions. Ensure necessary data like `DummyBeta.m`, `centEpiFibIC.m`, and `centBloodSub.m` are loaded.

```r
frac_cbs <- hepidish(
  beta.m = DummyBeta.m,
  ref1.m = centEpiFibIC.m,
  ref2.m = centBloodSub.m,
  h.CT.idx = 3,
  method = "CBS",
  nu.v = c(0.25, 0.5, 0.75)
)
```

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

### Inspect EpiDISH Output

Source: https://github.com/sjczheng/epidish/blob/devel/README.md

Examine the estimated fractions and the dimensions of the reference and data matrices used.

```R
out.l$estF
dim(out.l$ref)
dim(out.l$dataREF)
```

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

### Access and Construct DNAm Panels

Source: https://context7.com/sjczheng/epidish/llms.txt

Retrieve detailed probe selection results and construct custom DNA methylation panels using median summary methods.

```R
cd4t_probes <- panel_result$DetailedResults$CD4T
print(head(cd4t_probes))
```

```R
fractions_custom <- epidish(
  beta.m = new_sample_data,
  ref.m = custom_reference,
  method = "RPC"
)$estF
```

```R
panel_median <- ConstructDNAmPanel(
  dnam.matrix = dnam_matrix,
  cell.types.info = cell_info,
  panel.cell.types = c("CD4T", "Bcell", "Mono"),
  n.probes = 50,
  fdr.threshold = 0.01,
  summary.method = "median"
)
```

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

### Construct Custom DNAm Reference Panels

Source: https://context7.com/sjczheng/epidish/llms.txt

This snippet demonstrates how to use `ConstructDNAmPanel` to build custom cell-type-specific DNA methylation reference panels. It requires a DNA methylation matrix and cell type information for samples. The function selects top-ranked probes based on a specified metric.

```r
library(EpiDISH)

# Assume you have purified cell-type DNA methylation data
# dnam_matrix: beta values matrix (probes x samples), no NAs
# cell_info: data.frame with "CellType" column for each sample

# Example structure:
# dnam_matrix[1:5, 1:3]
#>              Sample1  Sample2  Sample3
#> cg00000029   0.521    0.534    0.498
#> cg00000108   0.123    0.145    0.089
#> cg00000165   0.867    0.854    0.891

# cell_info
#>           SampleID  CellType
#> Sample1   Sample1   CD4T
#> Sample2   Sample2   CD4T
#> Sample3   Sample3   Bcell

# Construct reference panel for specific cell types
panel_result <- ConstructDNAmPanel(
  dnam.matrix = dnam_matrix,             # Beta values (probes x samples)
  cell.types.info = cell_info,           # Must have "CellType" column
  panel.cell.types = c("CD4T", "Bcell", "Mono", "NK"),  # Target cell types
  n.probes = 100,                        # Top probes per cell type
  fdr.threshold = 0.05,                  # FDR for differential methylation
  p.adjust.method = "BH",                # Multiple testing correction
  summary.method = "mean",               # "mean" or "median"
  equal.variance = FALSE                 # For t-test
)

# Access the reference panel matrix (for use with epidish)
custom_reference <- panel_result$ReferencePanel
print(dim(custom_reference))
```

```r
#> [1] 400   4  # 100 probes x 4 cell types

print(head(custom_reference))
```

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

### Utilize Built-in Reference Matrices

Source: https://context7.com/sjczheng/epidish/llms.txt

Load pre-built DNA methylation reference matrices for blood and solid tissues to perform deconvolution.

```R
library(EpiDISH)

data(centDHSbloodDMC.m)
print(colnames(centDHSbloodDMC.m))

data(cent12CT.m)
print(colnames(cent12CT.m))

data(cent12CT450k.m)

data(centUniLIFE.m)

data(centEpiFibIC.m)
print(colnames(centEpiFibIC.m))

data(centBloodSub.m)

data(centEpiFibFatIC.m)
print(colnames(centEpiFibFatIC.m))

data(LiuDataSub.m)
blood_fractions <- epidish(LiuDataSub.m, cent12CT.m, method = "RPC")$estF

data(DummyBeta.m)
solid_fractions <- hepidish(
  DummyBeta.m,
  ref1.m = centEpiFibIC.m,         # Primary: Epi, Fib, IC
  ref2.m = centBloodSub.m,         # Secondary: immune subtypes
  h.CT.idx = 3,                    # Decompose IC (column 3)
  method = "RPC"
)
```

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

### Perform Meta-Analysis with DoMetaEfron

Source: https://context7.com/sjczheng/epidish/llms.txt

Execute meta-analysis across EWAS studies using empirical Bayes and Stouffer's method, including filtering and diagnostic plotting.

```R
library(EpiDISH)

meta_result <- DoMetaEfron(
  stat.m = stat_matrix,            # Signed statistics (t-stats)
  pval.m = pval_matrix,            # Associated p-values
  bre = 120,                       # Breakpoints for locfdr
  df = 15,                         # Degrees of freedom for spline
  pct0 = 0.25,                     # Percentage for fitting null (25%)
  plotlocfdr = 0                   # Set to 1 to see diagnostic plots
)

print(head(meta_result))

significant_cpgs <- rownames(meta_result)[meta_result[, "AdjP(BH)"] < 0.05]
length(significant_cpgs)

hyper_cpgs <- rownames(meta_result)[meta_result[, "z(Stouffer)"] > 0 &
                                     meta_result[, "AdjP(BH)"] < 0.05]
hypo_cpgs <- rownames(meta_result)[meta_result[, "z(Stouffer)"] < 0 &
                                    meta_result[, "AdjP(BH)"] < 0.05]

meta_with_plots <- DoMetaEfron(
  stat.m = stat_matrix,
  pval.m = pval_matrix,
  plotlocfdr = 1                   # Show locfdr diagnostic plots
)
```

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

### Estimate Breast Tissue Cell Fractions

Source: https://context7.com/sjczheng/epidish/llms.txt

Estimates cell-type fractions in breast tissue using the epidish function with the RPC method. Requires beta values and a reference matrix.

```r
breast_fractions <- epidish(
  breast_betas,
  centEpiFibFatIC.m,
  method = "RPC"
)$estF
```

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

### Infer Cell-Type Fractions with EpiDISH

Source: https://github.com/sjczheng/epidish/blob/devel/README.md

Use the epidish function with the RPC method to estimate cell-type fractions from beta values.

```R
out.l <- epidish(beta.m = DummyBeta.m, ref.m = centEpiFibIC.m, method = "RPC")
```

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

### Visualize inferred blood cell-type fractions

Source: https://github.com/sjczheng/epidish/blob/devel/README.md

Generates boxplots to visualize the distribution of inferred cell-type fractions, useful for validating results.

```R
boxplot(BloodFrac.m)
```

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

### Infer blood cell-type fractions using RPC

Source: https://github.com/sjczheng/epidish/blob/devel/README.md

Estimates cell-type fractions in a blood DNAm sample using the Robust Partial Correlations (RPC) method and a reference matrix.

```R
BloodFrac.m <- epidish(beta.m = LiuDataSub.m, ref.m = centDHSbloodDMC.m, method = "RPC")$estF
```

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

### Estimate Immune Cell Subtypes with HEpiDISH

Source: https://github.com/sjczheng/epidish/blob/devel/README.md

Perform hierarchical estimation of cell-type fractions using primary and secondary DNAm references.

```R
data(centBloodSub.m)
frac.m <- hepidish(beta.m = DummyBeta.m, ref1.m = centEpiFibIC.m, ref2.m = centBloodSub.m[,c(1, 2, 5)], h.CT.idx = 3, method = 'RPC')
frac.m
```

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

### Perform hierarchical deconvolution with hepidish

Source: https://context7.com/sjczheng/epidish/llms.txt

Uses hepidish to resolve cell-type fractions at multiple levels by decomposing a specific cell-type from a primary reference into subtypes from a secondary reference.

```r
library(EpiDISH)

# Load references for hierarchical deconvolution
data(centEpiFibIC.m)    # Primary: Epithelial, Fibroblast, Immune Cells
data(centBloodSub.m)    # Secondary: 7 immune cell subtypes
data(DummyBeta.m)       # Sample solid tissue data

# Perform hierarchical deconvolution
# h.CT.idx = 3 indicates "IC" (3rd column) in primary ref maps to secondary ref
frac_hierarchical <- hepidish(
  beta.m = DummyBeta.m,             # Beta values (CpGs x samples)
  ref1.m = centEpiFibIC.m,          # Primary reference (Epi, Fib, IC)
  ref2.m = centBloodSub.m,          # Secondary reference (immune subtypes)
  h.CT.idx = 3,                     # Index of cell-type to decompose (IC)
  method = "RPC",                   # Method: "RPC", "CBS", or "CP"
  maxit = 50
)

# Result contains: Epi, Fib, and 7 immune cell subtype fractions
print(frac_hierarchical)
#>              Epi       Fib    B-cells    CD4T     CD8T      NK    Mono   Neutro   Eosino
#> Sample1  0.4521   0.2134   0.0123   0.0456   0.0234   0.0167  0.0312  0.1834   0.0219
#> Sample2  0.3892   0.2567   0.0145   0.0523   0.0289   0.0198  0.0367  0.1756   0.0263

# Using only selected immune subtypes from secondary reference
frac_selected <- hepidish(
  beta.m = DummyBeta.m,
  ref1.m = centEpiFibIC.m,
  ref2.m = centBloodSub.m[, c("B", "NK", "Mono")],  # Subset of immune cells
  h.CT.idx = 3,
  method = "RPC"
)
```

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

### epidish - Cell-Type Fraction Estimation

Source: https://context7.com/sjczheng/epidish/llms.txt

The epidish function infers cell-type fractions from a DNA methylation beta-value matrix using a reference-based approach.

```APIDOC
## epidish

### Description
Infers cell-type fractions from a DNA methylation beta-value matrix using reference-based deconvolution methods (RPC, CBS, or CP).

### Parameters
- **beta.m** (matrix) - Required - Beta value matrix (CpGs x samples).
- **ref.m** (matrix) - Required - Reference centroids (CpGs x cell-types).
- **method** (string) - Optional - Deconvolution method: "RPC" (default), "CBS", or "CP".
- **maxit** (integer) - Optional - Maximum iterations for RPC method.
- **nu.v** (vector) - Optional - Candidate nu values for CBS method.
- **constraint** (string) - Optional - Constraint type for CP method: "inequality" (default) or "equality".

### Response
- **estF** (matrix) - Estimated cell-type fractions (samples x cell-types).
- **ref** (matrix) - Reference matrix subset used.
- **dataREF** (matrix) - Input data subset matching reference probes.
```

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

### View CellDMC Results

Source: https://github.com/sjczheng/epidish/blob/devel/README.md

Display the head of the DMCT prediction results.

```R
head(celldmc.o$dmct)
```

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

### Identify Differentially Methylated Cell-Types with CellDMC

Source: https://github.com/sjczheng/epidish/blob/devel/README.md

Run CellDMC using a phenotype vector and previously estimated cell-type fractions.

```R
pheno.v <- rep(c(0, 1), each = 5)
celldmc.o <- CellDMC(DummyBeta.m, pheno.v, frac.m)
```

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

### hepidish - Hierarchical Cell-Type Deconvolution

Source: https://context7.com/sjczheng/epidish/llms.txt

The hepidish function performs hierarchical deconvolution to estimate cell-type fractions at multiple levels of resolution.

```APIDOC
## hepidish

### Description
Implements hierarchical EpiDISH for multi-level cell-type resolution, using a primary reference for broad estimation and a secondary reference to decompose specific cell-types.

### Parameters
- **beta.m** (matrix) - Required - Beta values (CpGs x samples).
- **ref1.m** (matrix) - Required - Primary reference matrix.
- **ref2.m** (matrix) - Required - Secondary reference matrix for sub-decomposition.
- **h.CT.idx** (integer) - Required - Index of the cell-type in the primary reference to be decomposed by the secondary reference.
- **method** (string) - Optional - Deconvolution method: "RPC", "CBS", or "CP".
- **maxit** (integer) - Optional - Maximum iterations for RPC method.

### Response
- **fractions** (matrix) - Combined matrix of primary and decomposed cell-type fractions.
```

=== COMPLETE CONTENT === This response contains all available snippets from this library. No additional content exists. Do not make further requests.