diff --git a/NAMESPACE b/NAMESPACE
index ae935a0..1d4629c 100644
--- a/NAMESPACE
+++ b/NAMESPACE
@@ -16,6 +16,7 @@ export(eSparse)
 export(m3_select_path)
 export(pathAssign)
 export(plotBinTile)
+export(plotDiagnostics)
 export(plotIntersect)
 export(plotTrend)
 export(plotTrendCluster)
@@ -149,6 +150,7 @@ importFrom(stats,cor)
 importFrom(stats,cutree)
 importFrom(stats,dist)
 importFrom(stats,family)
+importFrom(stats,fitted)
 importFrom(stats,gaussian)
 importFrom(stats,glm)
 importFrom(stats,glm.control)
@@ -159,8 +161,12 @@ importFrom(stats,median)
 importFrom(stats,na.omit)
 importFrom(stats,p.adjust)
 importFrom(stats,poisson)
+importFrom(stats,predict)
+importFrom(stats,residuals)
+importFrom(stats,rstandard)
 importFrom(stats,sd)
 importFrom(stats,setNames)
+importFrom(stringr,str_remove)
 importFrom(stringr,str_remove_all)
 importFrom(stringr,str_split)
 importFrom(stringr,str_split_i)
diff --git a/R/ScMaSigProClass.R b/R/ScMaSigProClass.R
index 7fba472..af0f28d 100644
--- a/R/ScMaSigProClass.R
+++ b/R/ScMaSigProClass.R
@@ -36,6 +36,7 @@
 #' model.
 #' @slot mt_correction A character string specifying the p-value correction
 #' method.
+#' @slot link Type of link function to use in the model. Default is "log".
 #' @slot epsilon Model convergence tolerance.
 #' @slot selection_method A character string specifying the method for stepwise
 #' regression.
@@ -73,6 +74,7 @@ setClass(
     p_value = "numeric",
     min_na = "numeric",
     mt_correction = "character",
+    link = "character",
     epsilon = "numeric",
     offset = "logical",
     log_offset = "logical",
@@ -91,7 +93,7 @@ setClass(
       "bin_ptime_col", "path_prefix", "root_label", "ptime_col", "bin_method",
       "path_col", "bin_col", "bin_size_col", "bin_mem_col",
       "mt_correction", "selection_method", "anno_col", "cluster_method",
-      "use_dim", "fill_na"
+      "use_dim", "fill_na", "link"
     )
 
     for (slot_name in char_slots) {
@@ -146,6 +148,7 @@ setClass(
     bin_mem_col = "scmp_bin_members",
     mt_correction = "BH",
     selection_method = "backward",
+    link = "log",
     epsilon = 1e-8,
     offset = TRUE,
     log_offset = FALSE,
@@ -172,6 +175,7 @@ setClass(
 #' @slot assignment_matrix A matrix containing binary assignment to branching paths
 #' Additionally has two columns for assignment of binned Pseudotime and
 #' replicate.
+#' @slot offset A numeric vector containing the offset values.
 #'
 #' @name MatrixDesign
 #' @aliases MatrixDesign-class
@@ -183,12 +187,14 @@ setClass(
   representation(
     predictor_matrix = "matrix",
     groups.vector = "character",
-    assignment_matrix = "matrix"
+    assignment_matrix = "matrix",
+    offset = "numeric"
   ),
   prototype = list(
     predictor_matrix = matrix(NA, nrow = 0, ncol = 0),
     groups.vector = character(),
-    assignment_matrix = matrix(NA, nrow = 0, ncol = 0)
+    assignment_matrix = matrix(NA, nrow = 0, ncol = 0),
+    offset = numeric()
   ),
   validity = function(object) {
     if (!validObject(object@predictor_matrix)) {
@@ -200,6 +206,9 @@ setClass(
     if (!validObject(object@assignment_matrix)) {
       stop("assignment_matrix slot is not a valid matrix.")
     }
+    if (!validObject(object@offset)) {
+      stop("offset slot is not a valid vector.")
+    }
     TRUE
   }
 )
diff --git a/R/plotDiagnostics.R b/R/plotDiagnostics.R
new file mode 100644
index 0000000..f552c85
--- /dev/null
+++ b/R/plotDiagnostics.R
@@ -0,0 +1,235 @@
+#' @title Plot Model Diagnostics
+#'
+#' @description
+#' Plots model residuals for the requested gene. Can be used for model
+#' diagnostics of optimized model or full model.
+#'
+#' @importFrom stats residuals fitted rstandard predict
+#' @importFrom stringr str_remove
+#'
+#' @param scmpObj An object of class \code{\link{ScMaSigPro}}.
+#' @param feature_id Name of the gene to be plotted.
+#' @param model Type of model to be used for plotting. Can be either 'optimized',
+#' 'intercept'and 'full'. Default is 'optimized'.
+#'
+#' @return ggplot2 plot object.
+#'
+#' @author Priyansh Srivastava \email{spriyansh29@@gmail.com}
+#'
+#' @export
+#'
+plotDiagnostics <- function(scmpObj,
+                            feature_id,
+                            model = "optimized") {
+  # Check if the results exist
+  assert_that(!isEmpty(scmpObj@Significant@genes),
+    msg = paste("No Significant genes found, please run the workflow first")
+  )
+
+  # Check model type
+  assert_that(all(model %in% c("optimized", "full", "intercept")),
+    msg = paste("The requested gene is not available in the significant genes list")
+  )
+
+  # Get all the available genes
+  avail_genes <- unique(unlist(scmpObj@Significant@genes))
+
+  # Check if the requested gene is availble as singificant gene
+  assert_that(all(feature_id %in% avail_genes),
+    msg = paste("The requested gene is not available in the significant genes list")
+  )
+
+  # Get the counts for the genes
+  y_mat <- as.matrix(scmpObj@Dense@assays@data$bulk.counts)
+
+  # Extract gene counts
+  y_df <- y_mat[feature_id, , drop = TRUE] %>%
+    as.data.frame()
+  colnames(y_df) <- "y"
+
+  # Set theme for the plot
+  theme_set <- theme_minimal(base_size = 12) +
+    theme(
+      legend.position = "bottom",
+      panel.grid.major = element_line(
+        color = "grey90", linewidth = 0.3, linetype = "dashed"
+      ),
+      panel.grid.minor = element_blank()
+    )
+
+  # Based on model
+  if (model == "full" || model == "intercept") {
+    # Create Model input
+    regression_matrix <- scmpObj@Design@predictor_matrix
+  } else {
+    # Get coefficients
+    coeff_matrix <- showCoeff(scmpObj) %>%
+      data.frame()
+
+    # Subset
+    coeff_matrix_sub <- coeff_matrix[feature_id, , drop = FALSE]
+    all_terms <- colnames(coeff_matrix_sub)[-1]
+
+    # Remove any occurance of beta
+    all_terms <- str_remove(string = all_terms, pattern = "beta")
+
+    # Remove intercept
+    all_terms <- all_terms[-1]
+
+    # Get the terms
+    terms <- coeff_matrix_sub[apply(coeff_matrix_sub, 2, FUN = function(i) {
+      i != 0
+    })]
+
+    # Extract the prediction
+    prediction_matrix <- scmpObj@Design@predictor_matrix
+
+    # Create Model input
+    regression_matrix <- prediction_matrix[, all_terms, drop = FALSE]
+  }
+
+  # Combine regression matrix with counts
+  model_df <- cbind(y_df, regression_matrix)
+
+  # Extract offsets
+  if (scmpObj@Parameters@offset) {
+    offset_data <- scmpObj@Design@offset
+  } else {
+    offset_data <- NULL
+  }
+
+  if (model == "intercept") {
+    glm_s3 <- glm(y ~ 1,
+      data = as.data.frame(model_df),
+      family = scmpObj@Parameters@distribution,
+      epsilon = scmpObj@Parameters@epsilon,
+      offset = offset_data,
+      weights = NULL,
+      maxit = scmpObj@Parameters@max_it
+    )
+  } else {
+    # Fit Model
+    glm_s3 <- glm(y ~ .,
+      data = as.data.frame(model_df),
+      family = scmpObj@Parameters@distribution,
+      epsilon = scmpObj@Parameters@epsilon,
+      offset = offset_data,
+      weights = NULL,
+      maxit = scmpObj@Parameters@max_it
+    )
+  }
+  # Extract residuals
+  raw_residuals <- residuals(glm_s3, type = "response")
+  pearson_residuals <- residuals(glm_s3, type = "pearson")
+  deviance_residuals <- residuals(glm_s3, type = "deviance")
+  standardized_residuals <- rstandard(glm_s3)
+
+  # Extract fitted values
+  fitted_values <- fitted(glm_s3)
+
+  # Calculate predicted responses
+  predicted_responses <- predict(glm_s3, type = "response")
+
+  # Q-Q plot
+  # This plot of the standardized residuals allows you to check if the
+  # residuals are approximately normally distributed, which is an
+  # Calculate standardized residuals
+
+  # Create a dataframe for plotting
+  qq_df <- data.frame(std_residuals = standardized_residuals)
+
+  # Create the Normal Q-Q plot
+  qq_plot <- ggplot(qq_df, aes(sample = .data$std_residuals)) +
+    stat_qq(color = "#EE446F", size = 2) +
+    stat_qq_line(
+      color = "#15918A", alpha = 1,
+      linewidth = 1, linetype = "solid"
+    ) +
+    labs(
+      subtitle = feature_id,
+      title = "A. Normal Q-Q Plot of Standardized Residuals",
+      x = "Theoretical Quantiles",
+      y = "Standardized Residuals"
+    ) +
+    theme_set
+
+  # Residuals vs Fitted Plot:
+  # This plot helps to detect non-linearity, unequal error variances,
+  # and outliers. It is used to check the assumption of homoscedasticity.
+  # Assuming your glm model is stored in 'glm_s3'
+
+  # Create a dataframe for plotting
+  res_vs_fit_df <- data.frame(
+    Fitted = fitted_values,
+    StdResiduals = standardized_residuals
+  )
+
+  # Create the Residuals vs Fitted plot
+  stdRes_vs_fitted_plot <- ggplot(res_vs_fit_df, aes(x = .data$Fitted, y = .data$StdResiduals)) +
+    geom_point(color = "#EE446F", size = 2) + # Plot the points
+    geom_hline(yintercept = 0, linetype = "dashed", color = "#9F7BB8") + # Add a horizontal line at 0
+    labs(
+      subtitle = feature_id,
+      title = "B. Standardized Residuals vs Fitted Plot", x = "Fitted Values", y = "Standardized Residuals"
+    ) +
+    theme_set +
+    geom_smooth(
+      method = "loess", color = "#15918A", alpha = 0.4, formula = y ~ x,
+      linewidth = 1, linetype = "solid", se = TRUE, fill = "#FDC659"
+    ) # Add a loess smoothed line
+
+  # Scale-Location Plot (or Spread-Location Plot): This shows if residuals
+  # are spread equally along the ranges of predictors. This is a way to check
+  # the assumption of equal variance (homoscedasticity).
+  # Calculate square root of absolute standardized residuals
+  sqrt_abs_std_residuals <- sqrt(abs(standardized_residuals))
+
+  # Create a dataframe for plotting
+  scale_loc_df <- data.frame(
+    Fitted = fitted_values,
+    SqrtAbsStdResiduals = sqrt_abs_std_residuals
+  )
+
+  # Create the Scale-Location plot
+  scale_loc_plot <- ggplot(scale_loc_df, aes(x = .data$Fitted, y = .data$SqrtAbsStdResiduals)) +
+    geom_point(color = "#EE446F", size = 2) + # Plot the points
+    geom_smooth(
+      method = "loess", color = "#15918A", formula = y ~ x,
+      linewidth = 1, linetype = "solid", se = TRUE,
+      fill = "#FDC659", alpha = 0.4
+    ) +
+    labs(
+      subtitle = feature_id,
+      title = "C. Scale-Location Plot",
+      x = "Fitted Values",
+      y = "Sqrt of Absolute Standardized Residuals"
+    ) +
+    theme_set
+
+  # A Dfun plot is less standard and typically refers to a plot that can help
+  # assess the appropriateness of the link function in a GLM. This might involve
+  # plotting observed versus expected responses or derivatives of the link function.
+  # The exact implementation can depend on the specific family and link function used
+  # in your model. For a basic implementation, you might plot observed vs. predicted responses.
+
+  # Create a dataframe for plotting
+  dfun_df <- data.frame(Observed = model_df$y, Predicted = predicted_responses)
+
+  # Create the Dfun plot
+  defunc_plot <- ggplot(dfun_df, aes(x = .data$Observed, y = .data$Predicted)) +
+    geom_point(color = "#EE446F", size = 2) +
+    geom_abline(
+      intercept = 0, slope = 1, linetype = "solid", color = "#15918A",
+      linewidth = 1, alpha = 1
+    ) +
+    labs(
+      subtitle = feature_id,
+      title = "D. Observed vs Predicted",
+      x = "Observed",
+      y = "Predicted"
+    ) +
+    theme_set
+
+  # Return Plots
+  return((qq_plot + stdRes_vs_fitted_plot) / (scale_loc_plot + defunc_plot))
+}
diff --git a/R/sc.p.vector.R b/R/sc.p.vector.R
index 6951770..80627ac 100644
--- a/R/sc.p.vector.R
+++ b/R/sc.p.vector.R
@@ -17,6 +17,7 @@
 #' @param min_na Minimum values needed per gene across cells to estimate the
 #' model.
 #' @param family  Distribution of the error term.
+#' @param link Type of link function to use in the model. Default is "log".
 #' @param epsilon Model convergence tolerance.
 #' @param offset logical value specifying whether to use offset during fitting.
 #' @param log_offset A logical value specifying whether to take the logarithm of
@@ -48,7 +49,8 @@ sc.p.vector <- function(scmpObj, p_value = 0.05, mt_correction = "BH",
                         offset = TRUE,
                         parallel = FALSE,
                         log_offset = FALSE,
-                        max_it = 100) {
+                        max_it = 100,
+                        link = "log") {
   # Check the type of the 'design' parameter and set the corresponding variables
   assert_that(is(scmpObj, "ScMaSigPro"),
     msg = "Please provide object of class 'ScMaSigPro'"
@@ -64,6 +66,14 @@ sc.p.vector <- function(scmpObj, p_value = 0.05, mt_correction = "BH",
   dat <- dat[, as.character(rownames(dis))]
   G <- nrow(dat)
 
+  # Check for the log function
+  assert_that(link %in% c("log", "identity"),
+    msg = "link function should be either 'log' or 'identity'"
+  )
+
+  # Update the family
+  family[["link"]] <- link
+
   # Add check
   # assert_that((dat@Dim[1] > 1), msg = paste(min_na, "for 'min_na' is too high. Try lowering the threshold."))
   assert_that(min_na <= ncol(dat),
@@ -109,8 +119,9 @@ sc.p.vector <- function(scmpObj, p_value = 0.05, mt_correction = "BH",
     if (log_offset) {
       offsetData <- log(offsetData)
     }
+    scmpObj@Design@offset <- offsetData
   } else {
-    offsetData <- NULL
+    offsetData <- NULL # scmpObj@Design@offset
   }
 
   if (parallel) {
@@ -257,6 +268,7 @@ sc.p.vector <- function(scmpObj, p_value = 0.05, mt_correction = "BH",
     scmpObj@Parameters@mt_correction <- mt_correction
     scmpObj@Parameters@epsilon <- epsilon
     scmpObj@Parameters@distribution <- family
+    scmpObj@Parameters@link <- link
 
     return(scmpObj)
   }
diff --git a/R/sc.set.poly.R b/R/sc.set.poly.R
index 1c90f96..e8e23b6 100644
--- a/R/sc.set.poly.R
+++ b/R/sc.set.poly.R
@@ -90,11 +90,16 @@ sc.set.poly <- function(scmpObj,
     group.cols = c(3:ncol(com.cell.meta))
   )
 
+  # Create offset
+  offset_vector <- numeric(rep(nrow(designList$edesign)))
+  names(offset_vector) <- rownames(designList$edesign)
+
   # Create Object
   designObj <- new("MatrixDesign",
     predictor_matrix = as.matrix(designList$dis),
     groups.vector = designList$groups.vector,
-    assignment_matrix = as.matrix(designList$edesign)
+    assignment_matrix = as.matrix(designList$edesign),
+    offset = offset_vector
   )
 
   # Update Slot
diff --git a/R/sc.t.fit.R b/R/sc.t.fit.R
index 99c451a..6c65c6a 100644
--- a/R/sc.t.fit.R
+++ b/R/sc.t.fit.R
@@ -14,6 +14,7 @@
 #' regression.
 #' @param nvar_correction Argument for correcting significance level. See details.
 #' @param family Distribution of the error term.
+#' @param link Type of link function to use in the model. Default is "log".
 #' @param epsilon Model convergence tolerance.
 #' @param offset A logical value specifying whether to use offset during fitting.
 #' @param log_offset A logical value specifying whether to take the logarithm of
@@ -47,11 +48,20 @@ sc.t.fit <- function(scmpObj,
                      verbose = TRUE,
                      parallel = FALSE,
                      log_offset = scmpObj@Parameters@log_offset,
-                     max_it = scmpObj@Parameters@max_it) {
+                     max_it = scmpObj@Parameters@max_it,
+                     link = scmpObj@Parameters@link) {
   assert_that(is(scmpObj, "ScMaSigPro"),
     msg = "Please provide object of class 'ScMaSigPro'"
   )
 
+  # Check for the log function
+  assert_that(link %in% c("log", "identity"),
+    msg = "link function should be either 'log' or 'identity'"
+  )
+
+  # Update the family
+  family[["link"]] <- link
+
   # Transfer Data
   dis <- scmpObj@Design@predictor_matrix
   p_value <- scmpObj@Parameters@p_value
@@ -116,8 +126,9 @@ sc.t.fit <- function(scmpObj,
     if (log_offset) {
       offsetData <- log(offsetData)
     }
+    scmpObj@Design@offset <- offsetData
   } else {
-    offsetData <- NULL
+    offsetData <- NULL # scmpObj@Design@offset
   }
 
   if (parallel) {
@@ -509,6 +520,7 @@ sc.t.fit <- function(scmpObj,
   scmpObj@Parameters@epsilon <- epsilon
   scmpObj@Parameters@selection_method <- selection_method
   scmpObj@Parameters@distribution <- family
+  scmpObj@Parameters@link <- link
 
   return(scmpObj)
 }
diff --git a/data/scmp.ob.RData b/data/scmp.ob.RData
index 2ea2b76..b830ccb 100644
Binary files a/data/scmp.ob.RData and b/data/scmp.ob.RData differ
diff --git a/man/MatrixDesign-class.Rd b/man/MatrixDesign-class.Rd
index e5a8277..9978ad3 100644
--- a/man/MatrixDesign-class.Rd
+++ b/man/MatrixDesign-class.Rd
@@ -20,6 +20,8 @@ term of the polynomial GLM.}
 \item{\code{assignment_matrix}}{A matrix containing binary assignment to branching paths
 Additionally has two columns for assignment of binned Pseudotime and
 replicate.}
+
+\item{\code{offset}}{A numeric vector containing the offset values.}
 }}
 
 \keyword{classes}
diff --git a/man/ParameterConfig-class.Rd b/man/ParameterConfig-class.Rd
index baed751..da893a0 100644
--- a/man/ParameterConfig-class.Rd
+++ b/man/ParameterConfig-class.Rd
@@ -53,6 +53,8 @@ model.}
 \item{\code{mt_correction}}{A character string specifying the p-value correction
 method.}
 
+\item{\code{link}}{Type of link function to use in the model. Default is "log".}
+
 \item{\code{epsilon}}{Model convergence tolerance.}
 
 \item{\code{selection_method}}{A character string specifying the method for stepwise
diff --git a/man/plotDiagnostics.Rd b/man/plotDiagnostics.Rd
new file mode 100644
index 0000000..3c55073
--- /dev/null
+++ b/man/plotDiagnostics.Rd
@@ -0,0 +1,26 @@
+% Generated by roxygen2: do not edit by hand
+% Please edit documentation in R/plotDiagnostics.R
+\name{plotDiagnostics}
+\alias{plotDiagnostics}
+\title{Plot Model Diagnostics}
+\usage{
+plotDiagnostics(scmpObj, feature_id, model = "optimized")
+}
+\arguments{
+\item{scmpObj}{An object of class \code{\link{ScMaSigPro}}.}
+
+\item{feature_id}{Name of the gene to be plotted.}
+
+\item{model}{Type of model to be used for plotting. Can be either 'optimized',
+'intercept'and 'full'. Default is 'optimized'.}
+}
+\value{
+ggplot2 plot object.
+}
+\description{
+Plots model residuals for the requested gene. Can be used for model
+diagnostics of optimized model or full model.
+}
+\author{
+Priyansh Srivastava \email{spriyansh29@gmail.com}
+}
diff --git a/man/sc.p.vector.Rd b/man/sc.p.vector.Rd
index 000dc1e..1482286 100644
--- a/man/sc.p.vector.Rd
+++ b/man/sc.p.vector.Rd
@@ -15,7 +15,8 @@ sc.p.vector(
   offset = TRUE,
   parallel = FALSE,
   log_offset = FALSE,
-  max_it = 100
+  max_it = 100,
+  link = "log"
 )
 }
 \arguments{
@@ -45,6 +46,8 @@ model.}
 the offsets.}
 
 \item{max_it}{Maximum number of iterations to fit the model.}
+
+\item{link}{Type of link function to use in the model. Default is "log".}
 }
 \value{
 An object of class \code{\link{ScMaSigPro}}, with updated `Profile`
diff --git a/man/sc.t.fit.Rd b/man/sc.t.fit.Rd
index 6403b2e..e080f16 100644
--- a/man/sc.t.fit.Rd
+++ b/man/sc.t.fit.Rd
@@ -15,7 +15,8 @@ sc.t.fit(
   verbose = TRUE,
   parallel = FALSE,
   log_offset = scmpObj@Parameters@log_offset,
-  max_it = scmpObj@Parameters@max_it
+  max_it = scmpObj@Parameters@max_it,
+  link = scmpObj@Parameters@link
 )
 }
 \arguments{
@@ -43,6 +44,8 @@ regression.}
 the offsets.}
 
 \item{max_it}{Maximum number of iterations to fit the model.}
+
+\item{link}{Type of link function to use in the model. Default is "log".}
 }
 \value{
 An object of class \code{\link{ScMaSigPro}}, with updated `Estimate`
diff --git a/tests/testthat/test-reproduce-masigpro.R b/tests/testthat/test-reproduce-masigpro.R
index 141b7a0..0174501 100644
--- a/tests/testthat/test-reproduce-masigpro.R
+++ b/tests/testthat/test-reproduce-masigpro.R
@@ -85,12 +85,12 @@ test_that("Reproduce Results of MaSigPro", {
 
   # Step-7: Run sc.p.vector
   test.scmp.1 <- sc.p.vector(test.scmp.1,
-    min_na = 20, verbose = FALSE,
+    min_na = 20, verbose = FALSE, link = "identity",
     offset = FALSE, parallel = FALSE, max_it = 25,
     epsilon = 0.00001, family = gaussian()
   )
   test.scmp.3 <- sc.p.vector(test.scmp.3,
-    min_na = 20, verbose = FALSE,
+    min_na = 20, verbose = FALSE, link = "identity",
     offset = FALSE, parallel = FALSE, max_it = 25,
     epsilon = 0.00001, family = gaussian()
   )
diff --git a/vignettes/S4_Class_object.png b/vignettes/S4_Class_object.png
deleted file mode 100644
index 827216f..0000000
Binary files a/vignettes/S4_Class_object.png and /dev/null differ
diff --git a/vignettes/scMaSigPro-Class.Rmd b/vignettes/scMaSigPro-Class.Rmd
index 4f51ebf..22b040d 100644
--- a/vignettes/scMaSigPro-Class.Rmd
+++ b/vignettes/scMaSigPro-Class.Rmd
@@ -78,7 +78,7 @@ scmp.ob
 ## Slots of the scMaSigPro object
 
 ```{r, echo=FALSE}
-knitr::include_graphics("S4_Class_object.png")
+knitr::include_graphics("https://www.metapriyansh.com/scMaSigPro/imgs/S4_Class_object.png")
 ```
 
 ### `Sparse` \& `Dense` Slots
@@ -292,3 +292,4 @@ to other vignettes for more in-depth analysis.
 ```{r, "Session Info"}
 sessionInfo(package = "scMaSigPro")
 ```
+
diff --git a/vignettes/scMaSigPro-maSigPro.Rmd b/vignettes/scMaSigPro-maSigPro.Rmd
index 8fac00b..f84a57d 100644
--- a/vignettes/scMaSigPro-maSigPro.Rmd
+++ b/vignettes/scMaSigPro-maSigPro.Rmd
@@ -231,6 +231,7 @@ gc <- capture_output(fit <- p.vector(data.abiotic, design,
 scmp_ob <- sc.p.vector(scmp_ob,
   min_na = 20,
   verbose = FALSE,
+  link = "identity",
   offset = FALSE,
   max_it = 25,
   epsilon = 0.00001,