fix: Handle zero-sum samples in generate_taxa_test_single

- Add check for samples with zero total counts after filtering - Remove samples with zero counts before running linda analysis - Prevent NA values during TSS normalization - Fix Issue #69 Co-Authored-By: jrotzetter <[email protected]>
cafferychen777 · Nov 8, 2024 · 0e7f17f · 0e7f17f
1 parent 98f7c0d
commit 0e7f17f
Showing 1 changed file with 7 additions and 0 deletions.
diff --git a/R/generate_taxa_test_single.R b/R/generate_taxa_test_single.R
@@ -150,6 +150,13 @@ generate_taxa_test_single <- function(data.obj,
       mStat_filter(prev.filter = prev.filter,
                    abund.filter = abund.filter)
 
+    # Add this check before linda analysis
+    if (any(colSums(otu_tax_agg_filter) == 0)) {
+      keep_samples <- colSums(otu_tax_agg_filter) > 0
+      otu_tax_agg_filter <- otu_tax_agg_filter[, keep_samples]
+      meta_tab <- meta_tab[keep_samples, ]
+    }
+
     # Perform LinDA (Linear models for Differential Abundance) analysis
     linda.obj <- linda(
       feature.dat = otu_tax_agg_filter,