update icescreen v1.1.1 (#5165)

* update icescreen v1.1.1 * added test for multi genbank file
galaxyproject · Feb 28, 2023 · a9bc153 · a9bc153
1 parent 08ec371
commit a9bc153
Show file tree

Hide file tree

Showing 4 changed files with 1,831 additions and 3 deletions.
diff --git a/tools/icescreen/icescreen.xml b/tools/icescreen/icescreen.xml
@@ -1,7 +1,7 @@
 <tool id="icescreen" name="ICEscreen" version="@TOOL_VERSION@+galaxy@VERSION_SUFFIX@" profile="20.09">
     <description>detects and annotates ICEs (Integrative and Conjugative Elements) and IMEs (Integrative and Mobilizable Elements) in Firmicutes genomes.</description>
     <macros>
-        <token name="@TOOL_VERSION@">1.1.0</token>
+        <token name="@TOOL_VERSION@">1.1.1</token>
         <token name="@VERSION_SUFFIX@">0</token>
     </macros>
     <requirements>
@@ -45,7 +45,7 @@
         ]]>
     </command>
     <inputs>
-        <param name="genome" type="data" format="genbank" label="Input genomes to analyze in Genbank format" help="Each Genbank file can only contain ONE record, multi-genbank files are not supported. Each Genbank file must include the ORIGIN nucleotide sequence at the end"/>
+        <param name="genome" type="data" format="genbank" label="Input genomes to analyze in Genbank format" help="Multi-genbank files (i.e. gbff) are supported. Each record must include the ORIGIN nucleotide sequence at the end."/>
         <section name="taxonomy" title="Taxonomy of the genomes to analyse" expanded="False">
             <param name="mode" type="select" label="Firmicutes is the defaut parameter" help="Streptomyces is EXPERIMENTAL and is NOT recommended.">
                 <option value="firmicutes" selected="true">Firmicutes</option>
@@ -115,7 +115,6 @@
         <test expect_num_outputs="4">
             <param name="genome" value="genbank/NC_004668_137848_164286.gb" ftype="genbank" />
             <param name="optional_files" value="output_zip_all_files" />
-
             <output name="outzip" >
                 <assert_contents>
                     <has_archive_member path=".*/*_detected_SP_withMEIds.tsv"><has_text text="ICE_IME_id" /></has_archive_member>
@@ -156,6 +155,18 @@
                 </assert_contents>
             </output>
         </test>
+        <test expect_num_outputs="3">
+            <param name="genome" value="genbank/NC_013798_298468_322494_NC_020450_643089_661957.gb" ftype="genbank" />
+            <output name="summary" file="NC_013798_298468_322494_NC_020450_643089_661957_detected_ME.summary" ftype="txt" />
+            <output name="detected_me" file="NC_013798_298468_322494_NC_020450_643089_661957_detected_ME.tsv" ftype="tabular" />
+            <output name="detected_sp" >
+                <assert_contents>
+                    <has_text text="ICE_IME_id" />
+                    <has_text text="WP_044555479.1" />
+                    <has_text text="WP_015426013.1" />
+                </assert_contents>
+            </output>
+        </test>
     </tests>
     <help><![CDATA[
         .. class:: warningmark

diff --git a/...s/icescreen/test-data/NC_013798_298468_322494_NC_020450_643089_661957_detected_ME.summary b/...s/icescreen/test-data/NC_013798_298468_322494_NC_020450_643089_661957_detected_ME.summary
@@ -0,0 +1,82 @@
+
+##################### Main parameters of the ICEscreen analysis #####################
+Maximum distance in CDS between consecutive SPs within a segments:	100
+Maximum distance in CDS between consecutive SPs within an IME element:	14
+
+
+##################### ICEscreen statistics for genome accession NC_013798.1 #####################
+
+##### Mobile elements #####
+## Number of complete elements
+Complete ICE (4 types of SP):	1
+Complete IME (R+I, R+C+I with distance between consecutive SPs <= 14 CDSs):	0
+## Number of complete modules
+Conjugation module (R+C+V):	0
+Mobilizable element (R+C with distance between consecutive SPs <= 14 CDS):	0
+## Number of partial elements
+Partial ICE (at least V):	0
+Other partial element (R+C, R+V, V+C with distance between consecutive SPs > 14 CDS):	0
+## Composite elements
+Total nested elements (partial or complete):	0
+Host element:	0
+Guest element:	0
+Element that are both host and guest:	0
+
+##### Signature proteins #####
+## Total SPs detected
+Total SPs:	4
+Total Integrase:	1
+Total Relaxase:	1
+Total Coupling protein:	1
+Total VirB4:	1
+## Unassigned SPs
+Unassigned SPs:	0
+Unassigned Integrase:	0
+Unassigned Relaxase:	0
+Unassigned Coupling protein:	0
+Unassigned VirB4:	0
+
+##### Segments #####
+Number of segments with one element:	1
+Number of segments with several elements:	0
+Number of segments with nested elements:	0
+Number of segments with exclusively isolated SPs to manually verify:	0
+
+##################### ICEscreen statistics for genome accession NC_020450.1 #####################
+
+##### Mobile elements #####
+## Number of complete elements
+Complete ICE (4 types of SP):	1
+Complete IME (R+I, R+C+I with distance between consecutive SPs <= 14 CDSs):	0
+## Number of complete modules
+Conjugation module (R+C+V):	0
+Mobilizable element (R+C with distance between consecutive SPs <= 14 CDS):	0
+## Number of partial elements
+Partial ICE (at least V):	0
+Other partial element (R+C, R+V, V+C with distance between consecutive SPs > 14 CDS):	0
+## Composite elements
+Total nested elements (partial or complete):	0
+Host element:	0
+Guest element:	0
+Element that are both host and guest:	0
+
+##### Signature proteins #####
+## Total SPs detected
+Total SPs:	4
+Total Integrase:	1
+Total Relaxase:	1
+Total Coupling protein:	1
+Total VirB4:	1
+## Unassigned SPs
+Unassigned SPs:	0
+Unassigned Integrase:	0
+Unassigned Relaxase:	0
+Unassigned Coupling protein:	0
+Unassigned VirB4:	0
+
+##### Segments #####
+Number of segments with one element:	1
+Number of segments with several elements:	0
+Number of segments with nested elements:	0
+Number of segments with exclusively isolated SPs to manually verify:	0
+
diff --git a/tools/icescreen/test-data/NC_013798_298468_322494_NC_020450_643089_661957_detected_ME.tsv b/tools/icescreen/test-data/NC_013798_298468_322494_NC_020450_643089_661957_detected_ME.tsv
@@ -0,0 +1,3 @@
+ICE_IME_id	Segment_number	Genome_accession	Category_of_element	Category_of_integrase	Host_ICE_IME_ids	Guest_ICE_IME_ids	Colocalized_ICE_IME_ids	ICEline_format	ICE_consensus_superfamily_SP_conj_module	ICE_consensus_family_SP_conj_module	IME_relaxase_family_domains_blast	HMM_family_SP_conj_module	Integrase_upstream	Integrase_downstream	Relaxase	Coupling_protein	VirB4	List_SP_ordered_genomic_position	Other_potential_SP_conj_module_need_manual_curation_and_review	Other_potential_integrase_need_manual_curation_and_review	Comments_regarding_structure
+ID_1	1	NC_013798.1	Complete ICE (V+C+R+D)	one integrase DDE	-	-	-	V8.C7.R2.D	Tn5252	TnGBS2	-	C:VirD4, R:MOBP, V:VirB4	-	GALLO_RS01755	GALLO_RS01740	GALLO_RS01695	GALLO_RS01650	GALLO_RS01650, GALLO_RS01695, GALLO_RS01740, GALLO_RS01755	-	-	Downstream integrase GALLO_RS01755 has been associated to the structure ID_1 pre-merging because it is adjacent to the conj module and there is no upstream/downstream ambiguity. 
+ID_2	2	NC_020450.1	Complete ICE (C+R+V+I)	one integrase Tyr	-	-	-	C1.R1:V5.I	Tn916	-	-	C:TcpA, R:MOBT, V:VirB4	-	LILO_RS03350	LILO_RS03260	LILO_RS03250	LILO_RS03320	LILO_RS03250, LILO_RS03260, LILO_RS03320, LILO_RS03350	-	-	Downstream integrase LILO_RS03350 has been associated to the structure ID_2 pre-merging because it is adjacent to the conj module and there is no upstream/downstream ambiguity.