Find Pathway Enrichments

Find Pathway Enrichments

find_pathway_enrichments(
  mzroll_list,
  regression_significance,
  pathway_list,
  enrichment_method = "gsea",
  ranking_measure = "statistic",
  fdr_cutoff = 0.1,
  test_absolute_effects = TRUE,
  additional_grouping_vars = NULL
)

Arguments

mzroll_list

output of process_mzroll or process_mzroll_multi

  • features: one row per unique analyte (defined by a unique groupId),

  • samples: one row per unique sample (defined by a unique sampleId),

  • measurements: one row per peak (samples x peakgroups)
regression_significance

returned by diffex_mzroll; a tibble of tests performed.
pathway_list

a named list where names are pathways and entries are groupIds belonging to each pathway.
enrichment_method

method used to calculate pathway enrichments.

gsea

gene set enrichment analysis using fgsea

fisher

Fisher exact test to look for enrichment of differentially abundant metabolite among pathways.
ranking_measure

variable in regression_significance to use when calculating enrichment.
fdr_cutoff

minimum adjusted p-value to use for creating a the pathway summary plot.
test_absolute_effects

If TRUE then only consider the magnitude of test-statistics when calculating rankings for enrichment; if FALSE then consider sign as well.
additional_grouping_vars

sample-, or measurement-level variables to groupby when performing regression in addition to groupId.

Value

a list containing:

  • enrichment_table: a tibble containing each term's enrichment for every pathway and an enrichment_plot

  • enrichment_plots: pre-generated plots containing the most significant enrichments for each term

Examples


library(dplyr)

#> 
#> Attaching package: ‘dplyr’
#> The following objects are masked from ‘package:stats’:
#> 
#>     filter, lag
#> The following objects are masked from ‘package:base’:
#> 
#>     intersect, setdiff, setequal, union

regression_significance <- diffex_mzroll(
  nplug_mzroll_normalized,
  "normalized_log2_abundance",
  "limitation + limitation:DR + 0"
)

#> Warning: q-value calculation initially failed due to too many small p-values
#>         but claman was able to recover results
pathway_nest <- nplug_mzroll_normalized$features %>%
  dplyr::select(groupId, pathway) %>%
  tidyr::nest(pathway_members = groupId)

pathway_list <- purrr::map(
  pathway_nest$pathway_members,
  function(x) {
    as.character(x$groupId)
  }
)
names(pathway_list) <- pathway_nest$pathway

find_pathway_enrichments(
  mzroll_list = nplug_mzroll_normalized,
  regression_significance,
  pathway_list,
  test_absolute_effects = FALSE,
  enrichment_method = "fisher"
)

#> # A tibble: 100 × 9
#>    term   pathway estimate p.value conf.low conf.high method  alternative qvalue
#>    <chr>  <chr>      <dbl>   <dbl>    <dbl>     <dbl> <chr>   <chr>        <dbl>
#>  1 limit… Pentos…    3.85    0.174  0.585         Inf Fisher… greater      0.758
#>  2 limit… Pyrimi…    2.04    0.228  0.588         Inf Fisher… greater      0.758
#>  3 limit… Glycol…    2.49    0.207  0.582         Inf Fisher… greater      0.758
#>  4 limit… Lipid …    0.250   0.963  0.00837       Inf Fisher… greater      0.963
#>  5 limit… TCA cy…    0.380   0.961  0.0903        Inf Fisher… greater      0.963
#>  6 limit… Purine…    0.472   0.939  0.142         Inf Fisher… greater      0.963
#>  7 limit… Energe…    1.22    0.541  0.316         Inf Fisher… greater      0.963
#>  8 limit… Amino …    1.41    0.376  0.496         Inf Fisher… greater      0.940
#>  9 limit… Amino …    0.691   0.823  0.210         Inf Fisher… greater      0.963
#> 10 limit… Amino …    0.618   0.855  0.155         Inf Fisher… greater      0.963
#> # … with 90 more rows