From e29ba5ddbbac155a114fa0cfd1e05b1d49cc66be Mon Sep 17 00:00:00 2001
From: Chris Fields <cjfields@illinois.edu>
Date: Thu, 8 Feb 2024 16:10:43 -0600
Subject: [PATCH] fix an error with the tag, and address limits for derepFasta 
 that require limiting the number of processed reads to 100k

---
 main.nf                        | 2 +-
 templates/PerSampleDadaInfer.R | 4 ++--
 2 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/main.nf b/main.nf
index 1125ca2..c2050bd 100644
--- a/main.nf
+++ b/main.nf
@@ -732,7 +732,7 @@ if (params.reads != false || params.input != false ) { // TODO maybe we should c
     } else {
 
         process PerSampleInferDerepAndMerge {
-            tag { "PerSampleInferDerepAndMerge: ${meta.id}" }
+            tag { "PerSampleInferDerepAndMerge:${meta.id}" }
             publishDir "${params.outdir}/dada2-Derep-Single/Per-Sample", mode: "copy", overwrite: true
 
             input:
diff --git a/templates/PerSampleDadaInfer.R b/templates/PerSampleDadaInfer.R
index 28994ef..79fbdc2 100644
--- a/templates/PerSampleDadaInfer.R
+++ b/templates/PerSampleDadaInfer.R
@@ -11,13 +11,13 @@ if (!is.na(dadaOpt)) {
 cat("Processing:", "${meta.id}", "\\n")
 
 errF <- readRDS("errors.R1.RDS")
-derepF <- derepFastq("${reads[0]}")
+derepF <- derepFastq("${reads[0]}", n=100000)
 ddF <- dada(derepF, err=errF, multithread=${task.cpus}, pool=as.logical("${params.pool}"))
 saveRDS(ddF, "${meta.id}.dd.R1.RDS")
 
 if (file.exists ("errors.R2.RDS")) {
   errR <- readRDS("errors.R2.RDS")
-  derepR <- derepFastq("${reads[1]}")
+  derepR <- derepFastq("${reads[1]}", n=100000)
   ddR <- dada(derepR, err=errR, multithread=${task.cpus}, pool=as.logical("${params.pool}"))
   saveRDS(ddR, "${meta.id}.dd.R2.RDS")