Test of stratification of biological replicates based on tissue composition, one cell-type at the time, using Kaplan-meier curves.

test_stratification_cellularity() takes as input A `tbl` (with at least three columns for sample, feature and transcript abundance) or `SummarizedExperiment` (more convenient if abstracted to tibble with library(tidySummarizedExperiment)) and returns a consistent object (to the input) with additional columns for the statistics from the hypothesis test.

test_stratification_cellularity(
  .data,
  .formula,
  .sample = NULL,
  .transcript = NULL,
  .abundance = NULL,
  method = "cibersort",
  reference = X_cibersort,
  ...
)

# S4 method for spec_tbl_df
test_stratification_cellularity(
  .data,
  .formula,
  .sample = NULL,
  .transcript = NULL,
  .abundance = NULL,
  method = "cibersort",
  reference = X_cibersort,
  ...
)

# S4 method for tbl_df
test_stratification_cellularity(
  .data,
  .formula,
  .sample = NULL,
  .transcript = NULL,
  .abundance = NULL,
  method = "cibersort",
  reference = X_cibersort,
  ...
)

# S4 method for tidybulk
test_stratification_cellularity(
  .data,
  .formula,
  .sample = NULL,
  .transcript = NULL,
  .abundance = NULL,
  method = "cibersort",
  reference = X_cibersort,
  ...
)

# S4 method for SummarizedExperiment
test_stratification_cellularity(
  .data,
  .formula,
  .sample = NULL,
  .transcript = NULL,
  .abundance = NULL,
  method = "cibersort",
  reference = X_cibersort,
  ...
)

# S4 method for RangedSummarizedExperiment
test_stratification_cellularity(
  .data,
  .formula,
  .sample = NULL,
  .transcript = NULL,
  .abundance = NULL,
  method = "cibersort",
  reference = X_cibersort,
  ...
)

Arguments

.data

A `tbl` (with at least three columns for sample, feature and transcript abundance) or `SummarizedExperiment` (more convenient if abstracted to tibble with library(tidySummarizedExperiment))

.formula

A formula representing the desired linear model. The formula can be of two forms: multivariable (recommended) or univariable Respectively: \"factor_of_interest ~ .\" or \". ~ factor_of_interest\". The dot represents cell-type proportions, and it is mandatory. If censored regression is desired (coxph) the formula should be of the form \"survival::Surv\(y, dead\) ~ .\"

.sample

The name of the sample column

.transcript

The name of the transcript/gene column

.abundance

The name of the transcript/gene abundance column

method

A string character. Either \"cibersort\", \"epic\" or \"llsr\". The regression method will be chosen based on being multivariable: lm or cox-regression (both on logit-transformed proportions); or univariable: beta or cox-regression (on logit-transformed proportions). See .formula for multi- or univariable choice.

reference

A data frame. The transcript/cell_type data frame of integer transcript abundance

...

Further parameters passed to the method deconvolve_cellularity

Value

A consistent object (to the input) with additional columns for the statistics from the hypothesis test (e.g., log fold change, p-value and false discovery rate).

A consistent object (to the input) with additional columns for the statistics from the hypothesis test (e.g., log fold change, p-value and false discovery rate).

A consistent object (to the input) with additional columns for the statistics from the hypothesis test (e.g., log fold change, p-value and false discovery rate).

Details

`r lifecycle::badge("maturing")`

This routine applies a deconvolution method (e.g., Cibersort; DOI: 10.1038/nmeth.3337) and passes the proportions inferred into a generalised linear model (DOI:dx.doi.org/10.1007/s11749-010-0189-z) or a cox regression model (ISBN: 978-1-4757-3294-8)

Underlying method for the test: data deconvolve_cellularity( !!.sample, !!.transcript, !!.abundance, method=method, reference = reference, action="get", ... ) [..] mutate(.high_cellularity = .proportion > median(.proportion)) survival::survdiff(data = data, .my_formula)

Examples

library(dplyr)
library(tidyr)

tidybulk::se_mini |>
test_stratification_cellularity(
  survival::Surv(days, dead) ~ .,
  cores = 1
)
#> # A tibble: 22 × 6
#>    .cell_type                        cell_t…¹ .low_…² .high…³  pvalue plot      
#>    <chr>                             <list>     <dbl>   <dbl>   <dbl> <list>    
#>  1 cibersort.B.cells.naive           <tibble>    3.35   0.65   0.0389 <ggsrvplt>
#>  2 cibersort.B.cells.memory          <tibble>    3.35   0.65   0.0389 <ggsrvplt>
#>  3 cibersort.Plasma.cells            <tibble>   NA     NA     NA      <NULL>    
#>  4 cibersort.T.cells.CD8             <tibble>    3.22   0.783  0.299  <ggsrvplt>
#>  5 cibersort.T.cells.CD4.naive       <tibble>    2.43   1.57   0.502  <ggsrvplt>
#>  6 cibersort.T.cells.CD4.memory.res… <tibble>    2.77   1.23   0.782  <ggsrvplt>
#>  7 cibersort.T.cells.CD4.memory.act… <tibble>    3.8    0.2    0.0455 <ggsrvplt>
#>  8 cibersort.T.cells.follicular.hel… <tibble>    3.35   0.65   0.0389 <ggsrvplt>
#>  9 cibersort.T.cells.regulatory..Tr… <tibble>    3.22   0.783  0.774  <ggsrvplt>
#> 10 cibersort.T.cells.gamma.delta     <tibble>   NA     NA     NA      <NULL>    
#> # … with 12 more rows, and abbreviated variable names ¹​cell_type_proportions,
#> #   ²​.low_cellularity_expected, ³​.high_cellularity_expected