update default nextclade dataset tag to "2023-09-21T12:00:00Z" for al…

…l TheiaCov wfs (#208)

* update default nextclade dataset tag for SC2 to "2023-09-21T12:00:00Z". applied on wfs theiacov clearlabs, fasta, ilmn pe, ilmn se, ont

* KmerFinder to TheiaProk (#188)

* skeleton on kmerfinder task

* remove kmerfinder_db_name

* still trying to get kmerfinder to run

* is this working now?!?

* first working version of kmerfinder task

* parse top hit

* make results output file optional

* final version of task (for now..)

* add kmerfinder_bacteria to theiaprok!

* fix typo

* pass empty file for tests - kmerfinder_db

* fix input json

* add kmerfinder query coverage to theiaprok SE, PE, FASTA and ONT

* remove undeeded skip on kmerfinder (set to false by default)

* fix ops

* update md5sum (part1)

* update md5sum (part2)

* fiz sister typo

* missed a file

* update CI

* add kmerfinder_template_coverage to theiaprok output

* expose database name on kmerfinder outputs

* update md5sum

* fix bug - headers being outputed to datatable

* update default nextclade dataset tag for SC2 to "2023-09-21T12:00:00Z". applied on wfs theiacov clearlabs, fasta, ilmn pe, ilmn se, ont

* update CI

---------

Co-authored-by: Inês Mendes <[email protected]>

tests/workflows/theiacov/test_wf_theiacov_clearlabs.yml

@@ -243,7 +243,7 @@
     - path: miniwdl_run/call-nextclade/work/nextclade_dataset_dir/primers.csv
       md5sum: 5990c3483bf66ce607aeb90a44e7ef2e
     - path: miniwdl_run/call-nextclade/work/nextclade_dataset_dir/qc.json
-      md5sum: 6587a54553ad565d5f8ea7d214a797d4
+      md5sum: b01f4491a54941fea12ec5b04a10fb8c
     - path: miniwdl_run/call-nextclade/work/nextclade_dataset_dir/reference.fasta
       md5sum: c7ce05f28e4ec0322c96f24e064ef55c
     - path: miniwdl_run/call-nextclade/work/nextclade_dataset_dir/sequences.fasta

tests/workflows/theiacov/test_wf_theiacov_fasta.yml

@@ -38,7 +38,7 @@
       md5sum: 6808ca805661622ad65ae014a4b2a094
     - path: miniwdl_run/call-consensus_qc/work/_miniwdl_inputs/0/clearlabs.fasta.gz
     - path: miniwdl_run/call-nextclade/command
-      md5sum: b5ecaad831316b3bd8f066f1e71cc0a5
+      md5sum: ed29cde6f430eff4c408d9ea214ebe85
     - path: miniwdl_run/call-nextclade/inputs.json
     - path: miniwdl_run/call-nextclade/outputs.json
     - path: miniwdl_run/call-nextclade/stderr.txt
@@ -69,11 +69,11 @@
     - path: miniwdl_run/call-nextclade/work/nextclade_dataset_dir/reference.fasta
       md5sum: c7ce05f28e4ec0322c96f24e064ef55c
     - path: miniwdl_run/call-nextclade/work/nextclade_dataset_dir/sequences.fasta
-      md5sum: bb6b4e9e91304a396724bcb6344b8a5d
+      md5sum: ea475ab0a62a0a68fc3b1108fdff8a20
     - path: miniwdl_run/call-nextclade/work/nextclade_dataset_dir/tag.json
-      md5sum: 97e1309e683fbaaa839198d88cd4e2d9
+      md5sum: 6a17b1ee5449279af7bdd0922545d7b8
     - path: miniwdl_run/call-nextclade/work/nextclade_dataset_dir/tree.json
-      md5sum: 6892e6019bf88ec571b4560d66d3acb0
+      md5sum: 13eb330629b6ef17a070fcb6283bea2f
     - path: miniwdl_run/call-nextclade/work/nextclade_dataset_dir/virus_properties.json
     - path: miniwdl_run/call-nextclade/work/nextclade_gene_E.translation.fasta
       md5sum: dc43b1e98245a25c142aec52b29a07df

tests/workflows/theiacov/test_wf_theiacov_ont.yml

@@ -143,7 +143,7 @@
     - path: miniwdl_run/call-nextclade/work/nextclade_dataset_dir/primers.csv
       md5sum: 5990c3483bf66ce607aeb90a44e7ef2e
     - path: miniwdl_run/call-nextclade/work/nextclade_dataset_dir/qc.json
-      md5sum: 6587a54553ad565d5f8ea7d214a797d4
+      md5sum: b01f4491a54941fea12ec5b04a10fb8c
     - path: miniwdl_run/call-nextclade/work/nextclade_dataset_dir/reference.fasta
       md5sum: c7ce05f28e4ec0322c96f24e064ef55c
     - path: miniwdl_run/call-nextclade/work/nextclade_dataset_dir/sequences.fasta

workflows/theiacov/wf_theiacov_clearlabs.wdl

@@ -31,7 +31,7 @@ workflow theiacov_clearlabs {
     File? reference_genome
     # nextclade inputs
     String nextclade_dataset_reference = "MN908947"
-    String nextclade_dataset_tag = "2023-08-17T12:00:00Z"
+    String nextclade_dataset_tag = "2023-09-21T12:00:00Z"
     String? nextclade_dataset_name
     # kraken parameters
     String? target_org

workflows/theiacov/wf_theiacov_fasta.wdl

@@ -21,7 +21,7 @@ workflow theiacov_fasta {
     String input_assembly_method
     # nextclade inputs
     String nextclade_dataset_reference = "MN908947"
-    String nextclade_dataset_tag = "2023-08-17T12:00:00Z"
+    String nextclade_dataset_tag = "2023-09-21T12:00:00Z"
     String? nextclade_dataset_name
     # qc check parameters
     File? qc_check_table

workflows/theiacov/wf_theiacov_illumina_pe.wdl

@@ -44,7 +44,7 @@ workflow theiacov_illumina_pe {
     # nextclade inputs
     String nextclade_docker_image = "nextstrain/nextclade:2.14.0"
     String nextclade_dataset_reference = "MN908947"
-    String nextclade_dataset_tag = "2023-08-17T12:00:00Z"
+    String nextclade_dataset_tag = "2023-09-21T12:00:00Z"
     String? nextclade_dataset_name
     # nextclade flu inputs
     String nextclade_flu_h1n1_ha_tag = "2023-04-02T12:00:00Z"

workflows/theiacov/wf_theiacov_illumina_se.wdl

@@ -31,7 +31,7 @@ workflow theiacov_illumina_se {
     Int trim_window_size = 4
     # nextclade inputs
     String nextclade_dataset_reference = "MN908947"
-    String nextclade_dataset_tag = "2023-08-17T12:00:00Z"
+    String nextclade_dataset_tag = "2023-09-21T12:00:00Z"
     String? nextclade_dataset_name
     # reference values
     File? reference_genome

workflows/theiacov/wf_theiacov_ont.wdl

@@ -31,7 +31,7 @@ workflow theiacov_ont {
     Int min_length = 400
     # nextclade inputs
     String nextclade_dataset_reference = "MN908947"
-    String nextclade_dataset_tag = "2023-08-17T12:00:00Z"
+    String nextclade_dataset_tag = "2023-09-21T12:00:00Z"
     String? nextclade_dataset_name
     # reference values
     File? reference_genome