Add additional input checks to check for frequently occuring input erros

Author: browaeysrobin

R/muscat_de.R

@@ -86,20 +86,37 @@ perform_muscat_de_analysis = function(sce, sample_id, celltype_id, group_id, bat
     if(sum(is_make_names) != length(levels(SummarizedExperiment::colData(sce)[,celltype_id]))){
       stop("The levels of the factor SummarizedExperiment::colData(sce)[,celltype_id] should be a syntactically valid R names - see make.names")
     }
+  } else{
+    is_make_names = unique(sort(SummarizedExperiment::colData(sce)[,celltype_id])) == make.names(unique(sort(SummarizedExperiment::colData(sce)[,celltype_id])))
+    if(sum(is_make_names) != length(unique(sort((SummarizedExperiment::colData(sce)[,celltype_id]))))){
+      stop("All the cell type labels in SummarizedExperiment::colData(sce)[,celltype_id] should be syntactically valid R names - see make.names")
+    }
   }
+
   if(is.factor(SummarizedExperiment::colData(sce)[,group_id])){
     is_make_names = levels(SummarizedExperiment::colData(sce)[,group_id]) == make.names(levels(SummarizedExperiment::colData(sce)[,group_id]))
     if(sum(is_make_names) != length(levels(SummarizedExperiment::colData(sce)[,group_id]))){
       stop("The levels of the factor SummarizedExperiment::colData(sce)[,group_id] should be a syntactically valid R names - see make.names")
     }
+  } else{
+    is_make_names = unique(sort(SummarizedExperiment::colData(sce)[,group_id])) == make.names(unique(sort(SummarizedExperiment::colData(sce)[,group_id])))
+    if(sum(is_make_names) != length(unique(sort((SummarizedExperiment::colData(sce)[,group_id]))))){
+      stop("All the group/condition labels in SummarizedExperiment::colData(sce)[,group_id] should be syntactically valid R names - see make.names")
+    }
   }
   if(is.factor(SummarizedExperiment::colData(sce)[,sample_id])){
     is_make_names = levels(SummarizedExperiment::colData(sce)[,sample_id]) == make.names(levels(SummarizedExperiment::colData(sce)[,sample_id]))
     if(sum(is_make_names) != length(levels(SummarizedExperiment::colData(sce)[,sample_id]))){
       stop("The levels of the factor SummarizedExperiment::colData(sce)[,sample_id] should be a syntactically valid R names - see make.names")
     }
+  } else{
+    is_make_names = unique(sort(SummarizedExperiment::colData(sce)[,sample_id])) == make.names(unique(sort(SummarizedExperiment::colData(sce)[,sample_id])))
+    if(sum(is_make_names) != length(unique(sort((SummarizedExperiment::colData(sce)[,sample_id]))))){
+      stop("All the sample_id labels in SummarizedExperiment::colData(sce)[,sample_id] should be syntactically valid R names - see make.names")
+    }
   }
 
+  
   if(!is.character(contrasts)){
     stop("contrasts should be a character vector")
   }
@@ -173,9 +190,21 @@ perform_muscat_de_analysis = function(sce, sample_id, celltype_id, group_id, bat
                         sid = "id",   # sample IDs (ctrl/stim.1234)
                         drop = FALSE)  # drop all other SummarizedExperiment::colData columns ----------------- change to false
 
+  # test to see whether sample_ids are unique
+  if (sum(table(sce$sample_id, sce$group_id) %>% apply(1, function(row_oi){sum(row_oi > 0)}) > 1) > 0){
+    stop("One or more of your sample_ids belongs to more than one group/condition of interest. Please make sure that all sample_ids are uniquely divided over your groups/conditions.")
+  } 
+  
   pb = muscat::aggregateData(sce,
                              assay = assay_oi_pb, fun = fun_oi_pb,
                              by = c("cluster_id", "sample_id"))
+  
+  if(assay_oi_pb == "counts"){
+    libsizes = colSums(SummarizedExperiment::assay(pb))
+    if (!isTRUE(all(libsizes == floor(libsizes)))) {
+      warning("non-integer library sizes: are you sure you are working with raw counts?")
+    }
+  }
 
   # prepare the experiment info (ei) table if batches present
   if(length(batches) > 1){

R/pipeline_wrappers.R

@@ -33,6 +33,42 @@ get_abundance_expression_info = function(sce, sample_id, group_id, celltype_id,
 
   requireNamespace("dplyr")
   requireNamespace("ggplot2")
+  
+  # if some of these are factors, and not all levels have syntactically valid names - prompt to change this
+  if(is.factor(SummarizedExperiment::colData(sce)[,celltype_id])){
+    is_make_names = levels(SummarizedExperiment::colData(sce)[,celltype_id]) == make.names(levels(SummarizedExperiment::colData(sce)[,celltype_id]))
+    if(sum(is_make_names) != length(levels(SummarizedExperiment::colData(sce)[,celltype_id]))){
+      stop("The levels of the factor SummarizedExperiment::colData(sce)[,celltype_id] should be a syntactically valid R names - see make.names")
+    }
+  } else{
+    is_make_names = unique(sort(SummarizedExperiment::colData(sce)[,celltype_id])) == make.names(unique(sort(SummarizedExperiment::colData(sce)[,celltype_id])))
+    if(sum(is_make_names) != length(unique(sort((SummarizedExperiment::colData(sce)[,celltype_id]))))){
+      stop("All the cell type labels in SummarizedExperiment::colData(sce)[,celltype_id] should be syntactically valid R names - see make.names")
+    }
+  }
+  
+  if(is.factor(SummarizedExperiment::colData(sce)[,group_id])){
+    is_make_names = levels(SummarizedExperiment::colData(sce)[,group_id]) == make.names(levels(SummarizedExperiment::colData(sce)[,group_id]))
+    if(sum(is_make_names) != length(levels(SummarizedExperiment::colData(sce)[,group_id]))){
+      stop("The levels of the factor SummarizedExperiment::colData(sce)[,group_id] should be a syntactically valid R names - see make.names")
+    }
+  } else{
+    is_make_names = unique(sort(SummarizedExperiment::colData(sce)[,group_id])) == make.names(unique(sort(SummarizedExperiment::colData(sce)[,group_id])))
+    if(sum(is_make_names) != length(unique(sort((SummarizedExperiment::colData(sce)[,group_id]))))){
+      stop("All the group/condition labels in SummarizedExperiment::colData(sce)[,group_id] should be syntactically valid R names - see make.names")
+    }
+  }
+  if(is.factor(SummarizedExperiment::colData(sce)[,sample_id])){
+    is_make_names = levels(SummarizedExperiment::colData(sce)[,sample_id]) == make.names(levels(SummarizedExperiment::colData(sce)[,sample_id]))
+    if(sum(is_make_names) != length(levels(SummarizedExperiment::colData(sce)[,sample_id]))){
+      stop("The levels of the factor SummarizedExperiment::colData(sce)[,sample_id] should be a syntactically valid R names - see make.names")
+    }
+  } else{
+    is_make_names = unique(sort(SummarizedExperiment::colData(sce)[,sample_id])) == make.names(unique(sort(SummarizedExperiment::colData(sce)[,sample_id])))
+    if(sum(is_make_names) != length(unique(sort((SummarizedExperiment::colData(sce)[,sample_id]))))){
+      stop("All the sample_id labels in SummarizedExperiment::colData(sce)[,sample_id] should be syntactically valid R names - see make.names")
+    }
+  }
 
   ### Receiver abundance plots
 
@@ -218,6 +254,76 @@ get_abundance_expression_info_separate = function(sce_receiver, sce_sender, samp
   requireNamespace("dplyr")
   requireNamespace("ggplot2")
 
+  # if some of these are factors, and not all levels have syntactically valid names - prompt to change this
+  if(is.factor(SummarizedExperiment::colData(sce_receiver)[,celltype_id_receiver])){
+    is_make_names = levels(SummarizedExperiment::colData(sce_receiver)[,celltype_id_receiver]) == make.names(levels(SummarizedExperiment::colData(sce_receiver)[,celltype_id_receiver]))
+    if(sum(is_make_names) != length(levels(SummarizedExperiment::colData(sce_receiver)[,celltype_id_receiver]))){
+      stop("The levels of the factor SummarizedExperiment::colData(sce_receiver)[,celltype_id_receiver] should be a syntactically valid R names - see make.names")
+    }
+  } else{
+    is_make_names = unique(sort(SummarizedExperiment::colData(sce_receiver)[,celltype_id_receiver])) == make.names(unique(sort(SummarizedExperiment::colData(sce_receiver)[,celltype_id_receiver])))
+    if(sum(is_make_names) != length(unique(sort((SummarizedExperiment::colData(sce_receiver)[,celltype_id_receiver]))))){
+      stop("All the cell type labels in SummarizedExperiment::colData(sce_receiver)[,celltype_id_receiver] should be syntactically valid R names - see make.names")
+    }
+  }
+  
+  if(is.factor(SummarizedExperiment::colData(sce_receiver)[,group_id])){
+    is_make_names = levels(SummarizedExperiment::colData(sce_receiver)[,group_id]) == make.names(levels(SummarizedExperiment::colData(sce_receiver)[,group_id]))
+    if(sum(is_make_names) != length(levels(SummarizedExperiment::colData(sce_receiver)[,group_id]))){
+      stop("The levels of the factor SummarizedExperiment::colData(sce_receiver)[,group_id] should be a syntactically valid R names - see make.names")
+    }
+  } else{
+    is_make_names = unique(sort(SummarizedExperiment::colData(sce_receiver)[,group_id])) == make.names(unique(sort(SummarizedExperiment::colData(sce_receiver)[,group_id])))
+    if(sum(is_make_names) != length(unique(sort((SummarizedExperiment::colData(sce_receiver)[,group_id]))))){
+      stop("All the group/condition labels in SummarizedExperiment::colData(sce_receiver)[,group_id] should be syntactically valid R names - see make.names")
+    }
+  }
+  if(is.factor(SummarizedExperiment::colData(sce_receiver)[,sample_id])){
+    is_make_names = levels(SummarizedExperiment::colData(sce_receiver)[,sample_id]) == make.names(levels(SummarizedExperiment::colData(sce_receiver)[,sample_id]))
+    if(sum(is_make_names) != length(levels(SummarizedExperiment::colData(sce_receiver)[,sample_id]))){
+      stop("The levels of the factor SummarizedExperiment::colData(sce_receiver)[,sample_id] should be a syntactically valid R names - see make.names")
+    }
+  } else{
+    is_make_names = unique(sort(SummarizedExperiment::colData(sce_receiver)[,sample_id])) == make.names(unique(sort(SummarizedExperiment::colData(sce_receiver)[,sample_id])))
+    if(sum(is_make_names) != length(unique(sort((SummarizedExperiment::colData(sce_receiver)[,sample_id]))))){
+      stop("All the sample_id labels in SummarizedExperiment::colData(sce_receiver)[,sample_id] should be syntactically valid R names - see make.names")
+    }
+  }
+  # if some of these are factors, and not all levels have syntactically valid names - prompt to change this
+  if(is.factor(SummarizedExperiment::colData(sce_sender)[,celltype_id_sender])){
+    is_make_names = levels(SummarizedExperiment::colData(sce_sender)[,celltype_id_sender]) == make.names(levels(SummarizedExperiment::colData(sce_sender)[,celltype_id_sender]))
+    if(sum(is_make_names) != length(levels(SummarizedExperiment::colData(sce_sender)[,celltype_id_sender]))){
+      stop("The levels of the factor SummarizedExperiment::colData(sce_sender)[,celltype_id_sender] should be a syntactically valid R names - see make.names")
+    }
+  } else{
+    is_make_names = unique(sort(SummarizedExperiment::colData(sce_sender)[,celltype_id_sender])) == make.names(unique(sort(SummarizedExperiment::colData(sce_sender)[,celltype_id_sender])))
+    if(sum(is_make_names) != length(unique(sort((SummarizedExperiment::colData(sce_sender)[,celltype_id_sender]))))){
+      stop("All the cell type labels in SummarizedExperiment::colData(sce_sender)[,celltype_id_sender] should be syntactically valid R names - see make.names")
+    }
+  }
+  
+  if(is.factor(SummarizedExperiment::colData(sce_sender)[,group_id])){
+    is_make_names = levels(SummarizedExperiment::colData(sce_sender)[,group_id]) == make.names(levels(SummarizedExperiment::colData(sce_sender)[,group_id]))
+    if(sum(is_make_names) != length(levels(SummarizedExperiment::colData(sce_sender)[,group_id]))){
+      stop("The levels of the factor SummarizedExperiment::colData(sce_sender)[,group_id] should be a syntactically valid R names - see make.names")
+    }
+  } else{
+    is_make_names = unique(sort(SummarizedExperiment::colData(sce_sender)[,group_id])) == make.names(unique(sort(SummarizedExperiment::colData(sce_sender)[,group_id])))
+    if(sum(is_make_names) != length(unique(sort((SummarizedExperiment::colData(sce_sender)[,group_id]))))){
+      stop("All the group/condition labels in SummarizedExperiment::colData(sce_sender)[,group_id] should be syntactically valid R names - see make.names")
+    }
+  }
+  if(is.factor(SummarizedExperiment::colData(sce_sender)[,sample_id])){
+    is_make_names = levels(SummarizedExperiment::colData(sce_sender)[,sample_id]) == make.names(levels(SummarizedExperiment::colData(sce_sender)[,sample_id]))
+    if(sum(is_make_names) != length(levels(SummarizedExperiment::colData(sce_sender)[,sample_id]))){
+      stop("The levels of the factor SummarizedExperiment::colData(sce_sender)[,sample_id] should be a syntactically valid R names - see make.names")
+    }
+  } else{
+    is_make_names = unique(sort(SummarizedExperiment::colData(sce_sender)[,sample_id])) == make.names(unique(sort(SummarizedExperiment::colData(sce_sender)[,sample_id])))
+    if(sum(is_make_names) != length(unique(sort((SummarizedExperiment::colData(sce_sender)[,sample_id]))))){
+      stop("All the sample_id labels in SummarizedExperiment::colData(sce_sender)[,sample_id] should be syntactically valid R names - see make.names")
+    }
+  }
   ### Receiver plots and info
 
   metadata_abundance = SummarizedExperiment::colData(sce_receiver)[,c(sample_id, group_id, celltype_id_receiver)] %>% tibble::as_tibble()
@@ -567,18 +673,34 @@ get_DE_info = function(sce, sample_id, group_id, celltype_id, batches, covariate
     if(sum(is_make_names) != length(levels(SummarizedExperiment::colData(sce)[,celltype_id]))){
       stop("The levels of the factor SummarizedExperiment::colData(sce)[,celltype_id] should be a syntactically valid R names - see make.names")
     }
+  } else{
+    is_make_names = unique(sort(SummarizedExperiment::colData(sce)[,celltype_id])) == make.names(unique(sort(SummarizedExperiment::colData(sce)[,celltype_id])))
+    if(sum(is_make_names) != length(unique(sort((SummarizedExperiment::colData(sce)[,celltype_id]))))){
+      stop("All the cell type labels in SummarizedExperiment::colData(sce)[,celltype_id] should be syntactically valid R names - see make.names")
+    }
   }
+  
   if(is.factor(SummarizedExperiment::colData(sce)[,group_id])){
     is_make_names = levels(SummarizedExperiment::colData(sce)[,group_id]) == make.names(levels(SummarizedExperiment::colData(sce)[,group_id]))
     if(sum(is_make_names) != length(levels(SummarizedExperiment::colData(sce)[,group_id]))){
       stop("The levels of the factor SummarizedExperiment::colData(sce)[,group_id] should be a syntactically valid R names - see make.names")
     }
+  } else{
+    is_make_names = unique(sort(SummarizedExperiment::colData(sce)[,group_id])) == make.names(unique(sort(SummarizedExperiment::colData(sce)[,group_id])))
+    if(sum(is_make_names) != length(unique(sort((SummarizedExperiment::colData(sce)[,group_id]))))){
+      stop("All the group/condition labels in SummarizedExperiment::colData(sce)[,group_id] should be syntactically valid R names - see make.names")
+    }
   }
   if(is.factor(SummarizedExperiment::colData(sce)[,sample_id])){
     is_make_names = levels(SummarizedExperiment::colData(sce)[,sample_id]) == make.names(levels(SummarizedExperiment::colData(sce)[,sample_id]))
     if(sum(is_make_names) != length(levels(SummarizedExperiment::colData(sce)[,sample_id]))){
       stop("The levels of the factor SummarizedExperiment::colData(sce)[,sample_id] should be a syntactically valid R names - see make.names")
     }
+  } else{
+    is_make_names = unique(sort(SummarizedExperiment::colData(sce)[,sample_id])) == make.names(unique(sort(SummarizedExperiment::colData(sce)[,sample_id])))
+    if(sum(is_make_names) != length(unique(sort((SummarizedExperiment::colData(sce)[,sample_id]))))){
+      stop("All the sample_id labels in SummarizedExperiment::colData(sce)[,sample_id] should be syntactically valid R names - see make.names")
+    }
   }
 
   if(!is.character(contrasts_oi)){

README.Rmd

@@ -95,7 +95,8 @@ When applying MultiNicheNet on datasets with many samples and cell types, it is
 
 ## Frequently recurring questions and issues
 
-* Even though it is stated in the vignettes, many reported issues arise because names of celltypes, groups/conditions, and/or samples are not syntactically valid. Before reporting your issue, make sure you satisfy this condition and other conditions described in the vignettes.
+* Even though it is stated in the vignettes, many reported issues arise because names of celltypes, groups/conditions, and/or samples are not syntactically valid. Before reporting your issue, make sure you satisfy this condition and other conditions described in the vignettes. In the latest version of MultiNicheNet, input checks are run to check this and give an understandable error message.
+* It is required that each sample is uniquely assigned to only one condition/group of interest. See the vignettes about paired and multifactorial analysis to see how to define your analysis input when you have multiple samples and conditions per patient. In the latest version of MultiNicheNet, input checks are run to check this and give an understandable error message.
 * We strongly recommend having at least 4 samples in each of the groups/conditions you want to compare. With less samples, the benefits of performing a pseudobulk-based DE analysis are less clear and non-multi-sample tools for differential cell-cell communication might be better alternatives. 
 
 ## References

README.md

@@ -190,7 +190,14 @@ plots; and 2) interpreting the results and generating visualizations.
     arise because names of celltypes, groups/conditions, and/or samples
     are not syntactically valid. Before reporting your issue, make sure
     you satisfy this condition and other conditions described in the
-    vignettes.
+    vignettes. In the latest version of MultiNicheNet, input checks are
+    run to check this and give an understandable error message.
+-   It is required that each sample is uniquely assigned to only one
+    condition/group of interest. See the vignettes about paired and
+    multifactorial analysis to see how to define your analysis input
+    when you have multiple samples and conditions per patient. In the
+    latest version of MultiNicheNet, input checks are run to check this
+    and give an understandable error message.
 -   We strongly recommend having at least 4 samples in each of the
     groups/conditions you want to compare. With less samples, the
     benefits of performing a pseudobulk-based DE analysis are less clear