Error in makeFunctionalPrediction() – "no rows to aggregate"

[msuchet]

Hi ! ’m running Tax4Fun2 (v1.1.5) on Ubuntu via WSL2. The runRefBlast() step completed successfully, but when I call makeFunctionalPrediction(), I get the following error:

Error in aggregate.data.frame(x = otu_table_reduced[, -1], by = list(otu_table_reduced[,  :
  no rows to aggregate

Context and setup

• R on Ubuntu 22.04 (via WSL2 on Windows 11)
• Tax4Fun2 version: 1.1.5 (downloaded from Zenodo: https://zenodo.org/records/10035668)
• Reference data: Tax4Fun2_ReferenceData_v2 (properly extracted and readable)
• Database mode: "Ref99NR"

Path & script used :

fasta_file <- "rep-seqs_filtered.fasta"
temp_folder <- "~/Tax4fun_test_ubuntu/Tax4Fun2_temp_results"
ref_path <- "~/Tax4fun_test_ubuntu/Tax4Fun2_ReferenceData_v2"
if (!dir.exists(temp_folder)) dir.create(temp_folder)
res_tax4fun <- runRefBlast(path_to_otus = fasta_file,
                           path_to_reference_data = ref_path,
                           path_to_temp_folder = temp_folder,
                           database_mode = "Ref99NR",
                           use_force = TRUE,
                           num_threads = 4)
list.files(temp_folder)
# "logfile1.txt"  "ref_blast.txt" -> output here = ok

blast_res <- read.table(file.path(temp_folder, "ref_blast.txt"), sep = "\t", header = FALSE)
head(read.table("Tax4Fun2_temp_results/ref_blast.txt", nrows = 5)$V1)
# "Cluster_1" "Cluster_2" "Cluster_3" "Cluster_6" "Cluster_8" -> corresponding to my ASVs identifications

res_functional <- makeFunctionalPrediction(
  path_to_otu_table = "/home/msuchet/Tax4fun_test_ubuntu/otu_table_tax4fun2_corrected.txt",
  path_to_reference_data = ref_path,
  path_to_temp_folder = temp_folder,
  database_mode = "Ref99NR",
  normalize_by_copy_number = TRUE,
  min_identity_to_reference = 0.95,  
  normalize_pathways = FALSE)         
#Using minimum idenity cutoff of 95% to nearest neighbor
#Error in aggregate.data.frame(x = otu_table_reduced[, -1], by = list(otu_table_reduced[,  : no rows to aggregate

I verified that:

• The OTU table and reference blast results exist and are readable.
• OTU names match between my table and the ref_blast.txt output.
• The min identity threshold isn't the problem.
• The same error occurred previously on Windows, so it might not be system-dependent.

otu <- read.table("otu_table_tax4fun2_corrected.txt", header = TRUE, row.names = 1)
blast_res <- read.table(file.path(temp_folder, "ref_blast.txt"), sep = "\t", header = FALSE)
table(rownames(otu) %in% blast_res$V1)
# TRUE: 662
setdiff(rownames(otu), blast_res$V1)
# character(0)
table(blast_res$V3 >= 95) # TRUE = 438 -> 438 ASV with % identity >95%

So the question is, what could cause the error "no rows to aggregate" when I try to use makeFunctionalPrediction()? How to fix it ?
Any help or clarification would be greatly appreciated. I can share detaisl or files if needed.

Best

Marine