diff --git a/main.nf b/main.nf
index 6ba27ae..41ea7f8 100644
--- a/main.nf
+++ b/main.nf
@@ -177,7 +177,7 @@ process cmsearch_orientations {
 
     """
     python3 -c "from Bio import SeqIO;import gzip;SeqIO.write(SeqIO.parse(gzip.open('${R1}', 'rt'), 'fastq'), 'R1.fa', 'fasta')"
-    cmsearch -E 10.0 --cpu 32 --hmmonly --noali --tblout scores.txt ${model} R1.fa
+    cmsearch -E 10.0 --cpu ${params.nproc} --hmmonly --noali --tblout scores.txt ${model} R1.fa
     split_reads.py --counts counts.csv --cmsearch scores.txt ${sampleid} ${R1} ${R2}
     """
 }
@@ -198,7 +198,7 @@ process vsearch_orientations {
 
     """
     python3 -c "from Bio import SeqIO;import gzip;SeqIO.write(SeqIO.parse(gzip.open('${R1}', 'rt'), 'fastq'), 'R1.fa', 'fasta')"
-    vsearch --usearch_global R1.fa --db ${library} --id 0.75 --query_cov 0.8 --strand both --threads 32 --top_hits_only --userfields query+qstrand --userout hits.tsv
+    vsearch --usearch_global R1.fa --db ${library} --id 0.75 --query_cov 0.8 --strand both --threads ${params.nproc} --top_hits_only --userfields query+qstrand --userout hits.tsv
     split_reads.py --counts counts.csv --vsearch hits.tsv ${sampleid} ${R1} ${R2}
     """
 }
@@ -271,7 +271,7 @@ process learn_errors {
 
 process dada_dereplicate {
     // NOTE: sequences in reverse orientation are reverse complemented to forward orientation for clustering
-    label 'med_cpu_mem'
+    label 'c5d_2xlarge'
 
     input:
         tuple val(sampleid), val(batch), val(orientation), path(R1), path(R2), path(model)
@@ -291,16 +291,17 @@ process dada_dereplicate {
 
     """
     dada2_dada.R ${R1} ${R2} \
+        --counts counts.csv \
+        --data dada.rds \
         --errors ${model} \
-        --sampleid ${sampleid} \
+        --nthreads ${params.nproc} \
         --orientation ${orientation} \
+        --overlaps overlaps.csv \
         --params ${dada_params} \
-        --data dada.rds \
+        --sampleid ${sampleid} \
         --seqtab seqtab.csv \
         --seqtab-r1 seqtab_r1.csv \
-        --seqtab-r2 seqtab_r2.csv \
-        --counts counts.csv \
-        --overlaps overlaps.csv
+        --seqtab-r2 seqtab_r2.csv
     get_unmerged.R dada.rds \
         --forward-seqs unmerged_F.fasta \
         --reverse-seqs unmerged_R.fasta
diff --git a/nextflow.config b/nextflow.config
index 6f8b954..2ced88a 100644
--- a/nextflow.config
+++ b/nextflow.config
@@ -9,6 +9,7 @@ params {
     // Docker image is hosted on github
     container = "ghcr.io/nhoffman/dada2-nf:2.0.3"
     min_reads = 1
+    nproc = 4
     work_dir = "work"
 }
 
@@ -32,6 +33,7 @@ profiles {
         }
         params {
             output = 'output'
+            nproc = 32
         }
         process {
             container = params.container
@@ -43,8 +45,11 @@ profiles {
             // no more than 2 forks at a time.
             maxForks = 2
         }
+        withLabel: med_cpu_mem {
+            maxForks = 2
+        }
         executor{
-            queueSize = 16
+            queueSize = 32
         }
     }
 
@@ -56,6 +61,7 @@ profiles {
         }
         params {
             output = 'output'
+            nproc = 32
         }
         process {
             container = params.container
@@ -74,6 +80,7 @@ profiles {
         resume = true  // null for no resume
         params {
             output = 'output'
+            nproc = 32
         }
         process {
             container = params.container
@@ -92,6 +99,10 @@ profiles {
 
     uw_batch {
         workDir = 's3://molmicro-data/nextflow-workdir/dada2-nf'
+        params {
+            output = 'output'
+            nproc = 32
+        }
         process {
             scratch = "/docker_scratch"
             queue = 'molmicro-queue'
@@ -130,6 +141,7 @@ profiles {
         }
         params {
             output = 'output'
+            nproc = 4
         }
         process {
             container = params.container
@@ -139,8 +151,14 @@ profiles {
             errorStrategy = 'retry'
             maxRetries = 3
         }
+        withLabel: c5d_2xlarge {
+            // Meant to match the AWS Batch cd5.9xlarge which has 32 cpus. For
+            // local execution cpus must be configured at the application and
+            // no more than 2 forks at a time.
+            maxForks = 4
+        }
         executor{
-            queueSize = 4
+            queueSize = 14
         }
     }
 
@@ -150,6 +168,7 @@ profiles {
         resume = true
         params {
             output = 'output'
+            nproc = 32
         }
         process {
             executor = 'awsbatch'