Add option to append current kSNP3 run to an existing tree (#122)

* save SNPs_all output file for appending new files to existing tree

* add option to pass snps_all file to append trees to an existing one

* fix dyslexia

tasks/phylogenetic_inference/task_ksnp3.wdl

@@ -7,6 +7,7 @@ task ksnp3 {
     String cluster_name
     Int kmer_size = 19
     String ksnp3_args = "" # add -ML to calculate a maximum likelihood tree or -NJ to calculate a neighbor-joining tree
+    File? previous_ksnp3_snps
     String docker_image = "quay.io/staphb/ksnp3:3.1"
     Int memory = 8
     Int cpu = 4
@@ -48,12 +49,21 @@ task ksnp3 {
   cat ksnp3_input.tsv
 
   # run ksnp3 on input assemblies
-  kSNP3 -in ksnp3_input.tsv -outdir ksnp3 -k ~{kmer_size} -core -vcf ~{ksnp3_args}
+  kSNP3 \
+    -in ksnp3_input.tsv \
+    -outdir ksnp3 \
+    -k ~{kmer_size} \
+    -core -vcf \
+    ~{'-SNPs_all ' + previous_ksnp3_snps} \
+    ~{ksnp3_args}
 
   # rename ksnp3 outputs with cluster name 
   # sometimes the core nwk and fasta outputs do not have content
   mv -v ksnp3/core_SNPs_matrix.fasta ksnp3/~{cluster_name}_core_SNPs_matrix.fasta
   mv -v ksnp3/tree.core.tre ksnp3/~{cluster_name}_core.nwk
+
+  # to allow appending new genomes to an existing tree, save the SNPs_all file
+  mv -v ksnp3/SNPs_all ksnp3/~{cluster_name}_SNPs_all
 
   if [ -s ksnp3/~{cluster_name}_core_SNPs_matrix.fasta ]; then # is the file not-empty?
     echo "The core SNP matrix was produced" | tee SKIP_SNP_DIST # then do NOT skip
@@ -82,6 +92,7 @@ task ksnp3 {
     File? ksnp3_ml_tree = "ksnp3/~{cluster_name}_ML.nwk"
     File? ksnp3_nj_tree = "ksnp3/~{cluster_name}_NJ.nwk"
     File number_snps = "ksnp3/COUNT_SNPs"
+    File ksnp3_snps_all = "ksnp3/~{cluster_name}_SNPs_all"
     File ksnp3_input = "ksnp3_input.tsv"
     String skip_core_snp_dists = read_string("SKIP_SNP_DIST")
     Array[File] ksnp_outs = glob("ksnp3/*")

workflows/phylogenetics/wf_ksnp3.wdl

@@ -69,6 +69,7 @@ workflow ksnp3_workflow {
     String ksnp3_snp_dists_version = pan_snp_dists.snp_dists_version
     File ksnp3_core_vcf = ksnp3_task.ksnp3_core_vcf
     String ksnp3_core_snp_matrix_status = ksnp3_task.skip_core_snp_dists
+    File ksnp3_snps = ksnp3_task.ksnp3_snps_all
     # ordered matrixes and reordered trees
     File? ksnp3_core_snp_matrix = core_reorder_matrix.ordered_matrix
     File? ksnp3_core_tree = core_reorder_matrix.tree