Hisat2-build: index construction w/ truly optional inputs (#3649)

* index construction mode dependent on optional inputs rather than memory availability * added tests for new functionality * updated tests to use appropriate test data * cleaned up test.yml to only contain BUILD outputs * removed unnecessary extract ss call * added : to empty metamaps. Reverted to sarscov2 test files --------- Co-authored-by: James A. Fellows Yates <jfy133@gmail.com>
nf-core · Jul 20, 2023 · f2f4883 · f2f4883
1 parent a174ece
commit f2f4883
Show file tree

Hide file tree

Showing 3 changed files with 78 additions and 9 deletions.
diff --git a/modules/nf-core/hisat2/build/main.nf b/modules/nf-core/hisat2/build/main.nf
@@ -37,9 +37,9 @@ process HISAT2_BUILD {
     def hisat2_build_memory = params.hisat2_build_memory ? (params.hisat2_build_memory as nextflow.util.MemoryUnit).toGiga() : 0
     if (avail_mem >= hisat2_build_memory) {
         log.info "[HISAT2 index build] At least ${hisat2_build_memory} GB available, so using splice sites and exons to build HISAT2 index"
-        extract_exons = "hisat2_extract_exons.py $gtf > ${gtf.baseName}.exons.txt"
-        ss = "--ss $splicesites"
-        exon = "--exon ${gtf.baseName}.exons.txt"
+        extract_exons = gtf ? "hisat2_extract_exons.py $gtf > ${gtf.baseName}.exons.txt" : ""
+        ss = splicesites ? "--ss $splicesites" : ""
+        exon = gtf ? "--exon ${gtf.baseName}.exons.txt" : ""
     } else {
         log.info "[HISAT2 index build] Less than ${hisat2_build_memory} GB available, so NOT using splice sites and exons to build HISAT2 index."
         log.info "[HISAT2 index build] Use --hisat2_build_memory [small number] to skip this check."

diff --git a/tests/modules/nf-core/hisat2/build_test/main.nf b/tests/modules/nf-core/hisat2/build_test/main.nf
@@ -15,3 +15,24 @@ workflow test_hisat2_build {
     HISAT2_EXTRACTSPLICESITES ( gtf )
     HISAT2_BUILD ( fasta, gtf, HISAT2_EXTRACTSPLICESITES.out.txt )
 }
+
+workflow test_hisat2_build_fasta_only {
+    fasta = [ [id:'genome'],
+              file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true)
+            ]
+    gtf   = [ [id:'genome'],
+              file(params.test_data['sarscov2']['genome']['genome_gtf'], checkIfExists: true)
+            ]
+    HISAT2_BUILD ( fasta, [[:],[]], [[:],[]] )
+}
+
+workflow test_hisat2_build_fasta_ss_only {
+    fasta = [ [id:'genome'],
+              file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true)
+            ]
+    gtf   = [ [id:'genome'],
+              file(params.test_data['sarscov2']['genome']['genome_gtf'], checkIfExists: true)
+            ]
+    HISAT2_EXTRACTSPLICESITES ( gtf )
+    HISAT2_BUILD ( fasta, [[:],[]], HISAT2_EXTRACTSPLICESITES.out.txt )
+}
diff --git a/tests/modules/nf-core/hisat2/build_test/test.yml b/tests/modules/nf-core/hisat2/build_test/test.yml
@@ -1,23 +1,71 @@
 - name: hisat2 build test_hisat2_build
-  command: nextflow run ./tests/modules/nf-core/hisat2/build_test -entry test_hisat2_build -c ./tests/config/nextflow.config -c ./tests/modules/nf-core/hisat2/build/nextflow.config
+  command: nextflow run ./tests/modules/nf-core/hisat2/build -entry test_hisat2_build -c ./tests/config/nextflow.config -c ./tests/modules/nf-core/hisat2/build/nextflow.config
   tags:
-    - hisat2
     - hisat2/build
+    - hisat2
   files:
-    - path: output/hisat2/genome.splice_sites.txt
+    - path: output/hisat2/hisat2/genome.1.ht2
+      md5sum: 057cfa8a22b97ee9cff4c8d342498803
+    - path: output/hisat2/hisat2/genome.2.ht2
+      md5sum: 47b153cd1319abc88dda532462651fcf
+    - path: output/hisat2/hisat2/genome.3.ht2
+      md5sum: 4ed93abba181d8dfab2e303e33114777
+    - path: output/hisat2/hisat2/genome.4.ht2
+      md5sum: c25be5f8b0378abf7a58c8a880b87626
     - path: output/hisat2/hisat2/genome.5.ht2
       md5sum: 91198831aaba993acac1734138c5f173
+    - path: output/hisat2/hisat2/genome.6.ht2
+      md5sum: 265e1284ce85686516fae5d35540994a
     - path: output/hisat2/hisat2/genome.7.ht2
       md5sum: 9013eccd91ad614d7893c739275a394f
+    - path: output/hisat2/hisat2/genome.8.ht2
+      md5sum: 33cdeccccebe80329f1fdbee7f5874cb
+    - path: output/hisat2/versions.yml
+
+- name: hisat2 build test_hisat2_build_fasta_only
+  command: nextflow run ./tests/modules/nf-core/hisat2/build -entry test_hisat2_build_fasta_only -c ./tests/config/nextflow.config -c ./tests/modules/nf-core/hisat2/build/nextflow.config
+  tags:
+    - hisat2/build
+    - hisat2
+  files:
     - path: output/hisat2/hisat2/genome.1.ht2
-      md5sum: 057cfa8a22b97ee9cff4c8d342498803
+      md5sum: 98284d11e09faba5be6caeacceae7b3c
     - path: output/hisat2/hisat2/genome.2.ht2
       md5sum: 47b153cd1319abc88dda532462651fcf
-    - path: output/hisat2/hisat2/genome.6.ht2
-      md5sum: 265e1284ce85686516fae5d35540994a
     - path: output/hisat2/hisat2/genome.3.ht2
       md5sum: 4ed93abba181d8dfab2e303e33114777
+    - path: output/hisat2/hisat2/genome.4.ht2
+      md5sum: c25be5f8b0378abf7a58c8a880b87626
+    - path: output/hisat2/hisat2/genome.5.ht2
+      md5sum: 4db21638bce5ab535147c14a8c5ed27b
+    - path: output/hisat2/hisat2/genome.6.ht2
+      md5sum: 265e1284ce85686516fae5d35540994a
+    - path: output/hisat2/hisat2/genome.7.ht2
+      md5sum: 9013eccd91ad614d7893c739275a394f
     - path: output/hisat2/hisat2/genome.8.ht2
       md5sum: 33cdeccccebe80329f1fdbee7f5874cb
+    - path: output/hisat2/versions.yml
+
+- name: hisat2 build test_hisat2_build_fasta_ss_only
+  command: nextflow run ./tests/modules/nf-core/hisat2/build -entry test_hisat2_build_fasta_ss_only -c ./tests/config/nextflow.config -c ./tests/modules/nf-core/hisat2/build/nextflow.config
+  tags:
+    - hisat2/build
+    - hisat2
+  files:
+    - path: output/hisat2/hisat2/genome.1.ht2
+      md5sum: 057cfa8a22b97ee9cff4c8d342498803
+    - path: output/hisat2/hisat2/genome.2.ht2
+      md5sum: 47b153cd1319abc88dda532462651fcf
+    - path: output/hisat2/hisat2/genome.3.ht2
+      md5sum: 4ed93abba181d8dfab2e303e33114777
     - path: output/hisat2/hisat2/genome.4.ht2
       md5sum: c25be5f8b0378abf7a58c8a880b87626
+    - path: output/hisat2/hisat2/genome.5.ht2
+      md5sum: 91198831aaba993acac1734138c5f173
+    - path: output/hisat2/hisat2/genome.6.ht2
+      md5sum: 265e1284ce85686516fae5d35540994a
+    - path: output/hisat2/hisat2/genome.7.ht2
+      md5sum: 9013eccd91ad614d7893c739275a394f
+    - path: output/hisat2/hisat2/genome.8.ht2
+      md5sum: 33cdeccccebe80329f1fdbee7f5874cb
+    - path: output/hisat2/versions.yml