Skip to contents

Calculate gene expression trends across time points

Source: R/MultiSampAnalysis.R
CalSampGeneTrend.Rd

Analyzes gene expression patterns across temporal samples using linear regression or Mfuzz clustering to identify genes with different trend patterns.

Usage

CalSampGeneTrend(
  STID_obj = NULL,
  loop_id = NULL,
  samp_grp_index = FALSE,
  meta_key = NULL,
  niche_key = NULL,
  group_by = NULL,
  assay_id = "Spatial",
  layer_id = "data",
  method = "fitting",
  fitting_paras = list(lm_min_slope = 2, lm_pval_cutoff = 0.2),
  mfuzz_paras = list(cluster_num = 10, min_acore = 0.3),
  gene_list = NULL,
  remove_genes = NULL,
  col = COLOR_LIST[["PALETTE_WHITE_BG"]],
  return_data = FALSE,
  grp_nm = NULL,
  dir_nm = "M4_CalSampGeneTrend"
)

Arguments

STID_obj

A STID object

loop_id

Character, multi-sample analysis identifier

samp_grp_index

Logical, whether to group by sample groups instead of individual sample IDs (default: FALSE)

meta_key

Character, metadata key for retrieving cell data

niche_key

Character, niche key for filtering niche cells (optional)

group_by

Character, column name for cell type grouping

assay_id

Character, assay name (default: "Spatial")

layer_id

Character, layer/slot name (default: "data")

method

Character, trend analysis method - "fitting" (linear/non-linear) or "mfuzz" (fuzzy clustering) (default: "fitting")

fitting_paras

List of parameters for fitting method:

  • lm_min_slope: Minimum slope threshold for significant linear trend

  • lm_pval_cutoff: P-value cutoff for linear regression significance

mfuzz_paras

List of parameters for Mfuzz method:

  • cluster_num: Number of clusters for Mfuzz (default: 10)

  • min_acore: Minimum membership score for gene inclusion (default: 0.3)

gene_list

Character vector, specific genes to analyze (default: NULL, uses all variable features)

remove_genes

Character vector, genes to exclude from analysis

col

Character vector, color palette for visualization

return_data

Logical, whether to return results (default: FALSE)

grp_nm

Character, group name for output (default: NULL, uses timestamp)

dir_nm

Character, directory name for output (default: "M4_CalSampGeneTrend")

Value

If return_data = TRUE, returns a list of gene trend results per cell type

Details

The fitting method uses linear regression to identify genes with significant monotonic trends (increasing/decreasing) and quadratic regression to classify concave/convex patterns. The Mfuzz method performs fuzzy clustering to group genes with similar temporal expression patterns.

Examples

if (FALSE) { # \dontrun{
# Linear regression based trend analysis
results <- CalSampGeneTrend(
  STID_obj = stid_object,
  loop_id = "time_series",
  group_by = "cell_type",
  method = "fitting",
  fitting_paras = list(lm_min_slope = 1.5, lm_pval_cutoff = 0.05)
)

# Mfuzz clustering based trend analysis
results <- CalSampGeneTrend(
  STID_obj = stid_object,
  loop_id = "time_series",
  group_by = "cell_type",
  method = "mfuzz",
  mfuzz_paras = list(cluster_num = 8, min_acore = 0.4)
)
} # }