Merge pull request #6342 from martenson/ucsc-bump-polish

UCSC tools version bump & polish
galaxyproject · Sep 22, 2024 · 303002d · 303002d
2 parents 081635c + 17811e9
commit 303002d
Show file tree

Hide file tree

Showing 22 changed files with 107 additions and 67 deletions.
diff --git a/tools/ucsc_tools/fasplit/.lint_skip b/tools/ucsc_tools/fasplit/.lint_skip
diff --git a/tools/ucsc_tools/fasplit/.shed.yml b/tools/ucsc_tools/fasplit/.shed.yml
@@ -3,7 +3,7 @@ owner: iuc
 description: faSplit is a tool to split a single FASTA file into several files
 long_description: faSplit is a tool to split a single FASTA file into several files
 homepage_url: http://hgdownload.cse.ucsc.edu/admin/exe/
-remote_repository_url: https://github.com/galaxyproject/tools-iuc/tree/master/tools/ucsc-tools/fasplit
+remote_repository_url: https://github.com/galaxyproject/tools-iuc/tree/main/tools/ucsc_tools/fasplit
 type: unrestricted
 categories:
   - Fasta Manipulation
diff --git a/tools/ucsc_tools/fasplit/fasplit.xml b/tools/ucsc_tools/fasplit/fasplit.xml
@@ -1,11 +1,11 @@
-<tool id="fasplit" name="faSplit" version="@TOOL_VERSION@">
+<tool id="fasplit" name="faSplit" version="@TOOL_VERSION@" profile="22.05">
     <description>Split a FASTA file</description>
+    <macros>
+        <token name="@TOOL_VERSION@">469</token>
+    </macros>
     <xrefs>
         <xref type="bio.tools">UCSC_Genome_Browser_Utilities</xref>
     </xrefs>
-    <macros>
-        <token name="@TOOL_VERSION@">377</token>
-    </macros>
     <requirements>
         <requirement type="package" version="@TOOL_VERSION@">ucsc-fasplit</requirement>
     </requirements>
@@ -57,61 +57,65 @@ out/
                 <element name="2" file="2.fa" />
                 <element name="3" file="3.fa" />
                 <element name="4" file="4.fa" />
-            </output_collection>    
+            </output_collection>
         </test>
         <test>
             <param name="input" ftype="fasta" value="CDS.fa"/>
             <param name="split_type" value="byname"/>
             <output_collection name="output_list" type="list">
                 <element name="ENSCAFT00000022939_canisfamiliaris" file="ENSCAFT00000022939_canisfamiliaris.fa" />
-                <element name="ENSMUST00000207100_musmusculus" file="ENSMUST00000207100_musmusculus.fa" />
-                <element name="ENSSSCT00000014817_susscrofa" file="ENSSSCT00000014817_susscrofa.fa" />
                 <element name="ENSCAFT00000022963_canisfamiliaris" file="ENSCAFT00000022963_canisfamiliaris.fa" />
-                <element name="ENSMUST00000208839_musmusculus" file="ENSMUST00000208839_musmusculus.fa" />
-                <element name="ENSSSCT00000023183_susscrofa" file="ENSSSCT00000023183_susscrofa.fa" />
                 <element name="ENSCAFT00000025950_canisfamiliaris" file="ENSCAFT00000025950_canisfamiliaris.fa" />
-                <element name="ENSPTRT00000019089_pantroglodytes" file="ENSPTRT00000019089_pantroglodytes.fa" />
-                <element name="ENSSSCT00000032764_susscrofa" file="ENSSSCT00000032764_susscrofa.fa" />
                 <element name="ENSMUST00000026013_musmusculus" file="ENSMUST00000026013_musmusculus.fa" />
-                <element name="ENSPTRT00000034846_pantroglodytes" file="ENSPTRT00000034846_pantroglodytes.fa" />
-                <element name="ENSSSCT00000033745_susscrofa" file="ENSSSCT00000033745_susscrofa.fa" />
                 <element name="ENSMUST00000040820_musmusculus" file="ENSMUST00000040820_musmusculus.fa" />
-                <element name="ENSPTRT00000040520_pantroglodytes" file="ENSPTRT00000040520_pantroglodytes.fa" />
-                <element name="ENSSSCT00000035258_susscrofa" file="ENSSSCT00000035258_susscrofa.fa" />
                 <element name="ENSMUST00000041588_musmusculus" file="ENSMUST00000041588_musmusculus.fa" />
-                <element name="ENSPTRT00000040521_pantroglodytes" file="ENSPTRT00000040521_pantroglodytes.fa" />
-                <element name="ENST00000302850_homosapiens" file="ENST00000302850_homosapiens.fa" />
                 <element name="ENSMUST00000091291_musmusculus" file="ENSMUST00000091291_musmusculus.fa" />
-                <element name="ENSRNOT00000041155_rattusnorvegicus" file="ENSRNOT00000041155_rattusnorvegicus.fa" />
-                <element name="ENST00000338702_homosapiens" file="ENST00000338702_homosapiens.fa" />
                 <element name="ENSMUST00000100505_musmusculus" file="ENSMUST00000100505_musmusculus.fa" />
-                <element name="ENSRNOT00000044009_rattusnorvegicus" file="ENSRNOT00000044009_rattusnorvegicus.fa" />
-                <element name="ENST00000340611_homosapiens" file="ENST00000340611_homosapiens.fa" />
                 <element name="ENSMUST00000110806_musmusculus" file="ENSMUST00000110806_musmusculus.fa" />
-                <element name="ENSRNOT00000064726_rattusnorvegicus" file="ENSRNOT00000064726_rattusnorvegicus.fa" />
-                <element name="ENST00000341500_homosapiens" file="ENST00000341500_homosapiens.fa" />
                 <element name="ENSMUST00000153440_musmusculus" file="ENSMUST00000153440_musmusculus.fa" />
-                <element name="ENSRNOT00000066674_rattusnorvegicus" file="ENSRNOT00000066674_rattusnorvegicus.fa" />
-                <element name="ENST00000378069_homosapiens" file="ENST00000378069_homosapiens.fa" />
                 <element name="ENSMUST00000163344_musmusculus" file="ENSMUST00000163344_musmusculus.fa" />
-                <element name="ENSRNOT00000067448_rattusnorvegicus" file="ENSRNOT00000067448_rattusnorvegicus.fa" />
-                <element name="ENST00000421712_homosapiens" file="ENST00000421712_homosapiens.fa" />
                 <element name="ENSMUST00000168613_musmusculus" file="ENSMUST00000168613_musmusculus.fa" />
-                <element name="ENSSSCT00000008295_susscrofa" file="ENSSSCT00000008295_susscrofa.fa" />
-                <element name="ENST00000542639_homosapiens" file="ENST00000542639_homosapiens.fa" />
                 <element name="ENSMUST00000173143_musmusculus" file="ENSMUST00000173143_musmusculus.fa" />
+                <element name="ENSMUST00000207100_musmusculus" file="ENSMUST00000207100_musmusculus.fa" />
+                <element name="ENSMUST00000208839_musmusculus" file="ENSMUST00000208839_musmusculus.fa" />
+                <element name="ENSPTRT00000019089_pantroglodytes" file="ENSPTRT00000019089_pantroglodytes.fa" />
+                <element name="ENSPTRT00000034846_pantroglodytes" file="ENSPTRT00000034846_pantroglodytes.fa" />
+                <element name="ENSPTRT00000040520_pantroglodytes" file="ENSPTRT00000040520_pantroglodytes.fa" />
+                <element name="ENSPTRT00000040521_pantroglodytes" file="ENSPTRT00000040521_pantroglodytes.fa" />
+                <element name="ENSRNOT00000041155_rattusnorvegicus" file="ENSRNOT00000041155_rattusnorvegicus.fa" />
+                <element name="ENSRNOT00000044009_rattusnorvegicus" file="ENSRNOT00000044009_rattusnorvegicus.fa" />
+                <element name="ENSRNOT00000064726_rattusnorvegicus" file="ENSRNOT00000064726_rattusnorvegicus.fa" />
+                <element name="ENSRNOT00000066674_rattusnorvegicus" file="ENSRNOT00000066674_rattusnorvegicus.fa" />
+                <element name="ENSRNOT00000067448_rattusnorvegicus" file="ENSRNOT00000067448_rattusnorvegicus.fa" />
+                <element name="ENSSSCT00000008295_susscrofa" file="ENSSSCT00000008295_susscrofa.fa" />
                 <element name="ENSSSCT00000013404_susscrofa" file="ENSSSCT00000013404_susscrofa.fa" />
+                <element name="ENSSSCT00000014817_susscrofa" file="ENSSSCT00000014817_susscrofa.fa" />
+                <element name="ENSSSCT00000023183_susscrofa" file="ENSSSCT00000023183_susscrofa.fa" />
+                <element name="ENSSSCT00000032764_susscrofa" file="ENSSSCT00000032764_susscrofa.fa" />
+                <element name="ENSSSCT00000033745_susscrofa" file="ENSSSCT00000033745_susscrofa.fa" />
+                <element name="ENSSSCT00000035258_susscrofa" file="ENSSSCT00000035258_susscrofa.fa" />
+                <element name="ENST00000302850_homosapiens" file="ENST00000302850_homosapiens.fa" />
+                <element name="ENST00000338702_homosapiens" file="ENST00000338702_homosapiens.fa" />
+                <element name="ENST00000340611_homosapiens" file="ENST00000340611_homosapiens.fa" />
+                <element name="ENST00000341500_homosapiens" file="ENST00000341500_homosapiens.fa" />
+                <element name="ENST00000378069_homosapiens" file="ENST00000378069_homosapiens.fa" />
+                <element name="ENST00000421712_homosapiens" file="ENST00000421712_homosapiens.fa" />
+                <element name="ENST00000542639_homosapiens" file="ENST00000542639_homosapiens.fa" />
                 <element name="ENST00000600492_homosapiens" file="ENST00000600492_homosapiens.fa" />
-            </output_collection>    
+            </output_collection>
         </test>
     </tests>
     <help><![CDATA[
 **What it does**
 
 `faSplit`_ is a tool to split a single FASTA file into several files.
 
+For implementation details see faSplit's `source code`_.
+
 .. _faSplit: http://hgdownload.cse.ucsc.edu/admin/exe/linux.x86_64/FOOTER.txt
+.. _source code: https://github.com/ucscGenomeBrowser/kent/blob/master/src/utils/faSplit/faSplit.c
     ]]></help>
     <citations>
+        <citation type="doi">10.1093/bib/bbs038</citation>
     </citations>
 </tool>
diff --git a/tools/ucsc_tools/fatovcf/.shed.yml b/tools/ucsc_tools/fatovcf/.shed.yml
@@ -3,7 +3,7 @@ owner: iuc
 description: Convert a FASTA alignment file to Variant Call Format (VCF) single-nucleotide diffs
 homepage_url: http://hgdownload.cse.ucsc.edu/admin/exe/
 long_description: Convert a FASTA alignment file to Variant Call Format (VCF) single-nucleotide diffs by comparing the given FASTA sequences with a selectable reference. In default setting the first entry of the FASTA alignment file will serve as a reference.
-remote_repository_url: https://github.com/galaxyproject/tools-iuc/tree/master/tools/ucsc-tools/fatovcf
+remote_repository_url: https://github.com/galaxyproject/tools-iuc/tree/main/tools/ucsc_tools/fatovcf
 type: unrestricted
 categories:
 - Convert Formats
diff --git a/tools/ucsc_tools/fatovcf/fatovcf.xml b/tools/ucsc_tools/fatovcf/fatovcf.xml
@@ -48,7 +48,7 @@
     ]]></command>
     <inputs>
         <param name="in_fasta" format="fasta" type="data" label="FASTA Alignment" help="Must contain a series of sequences with different names and the same length. Both N and - are treated as missing information." />
-        
+
         <conditional name="ref_seq">
             <param name="refSeq" type="select" label="Determine reference sequence" help="Which sequence from the FASTA file should be used as the reference sequence.">
                 <option value="" selected="true">Use the first sequence as reference</option>
@@ -76,7 +76,7 @@
             <param argument="-windowSize" type="integer" min="0" value="0" label="Window size to mask bases" help="Mask any base for which there are at least -minAmbigWindow bases in a window of +-N bases around the base. Masking approach adapted from https://github.com/roblanf/sarscov2phylo/blob/master/scripts/mask_seq.py Use -windowSize=7 for same results. Default: 0" />
             <param argument="-minAmbigInWindow" type="integer" min="1" value="2" label="Minimum of ambiguous characters within the window given above" help="When -windowSize is provided, mask any base for which there are at least this many N, ambiguous or gap characters within the window. Default: 2" />
         </section>
-        
+
         <param argument="-includeNoAltN" type="boolean" truevalue="-includeNoAltN" falsevalue="" label="Include positions without defined ALT allele" help="Include base positions with no alternate alleles observed, but at least one N (missing base/no-call). Default: false" />
         <param argument="-minAc" type="integer" min="0" value="0" label="Minimum allele count" help="Ignore alternate alleles observed fewer than N times. Default: 0" />
         <param argument="-minAf" type="float" min="0.0" max="1.0" value="0.0" label="Minimum allele frequency" help="Ignore alternate alleles observed in less than F of non-N bases. Default: 0.0" />
@@ -136,7 +136,11 @@
 
 `faToVcf`_ is a tool to extract a VCF from a multi-sequence FASTA alignment.
 
-.. _faToVcf: http://hgdownload.cse.ucsc.edu/admin/exe/linux.x86_64/FOOTER.txt 
+For implementation details see faToVcf's `source code`_.
+
+.. _faToVcf: http://hgdownload.cse.ucsc.edu/admin/exe/linux.x86_64/FOOTER.txt
+.. _source code: https://github.com/ucscGenomeBrowser/kent/blob/master/src/hg/utils/faToVcf/faToVcf.c
+
     ]]>    </help>
     <citations>
         <citation type="doi">10.1093/bib/bbs038</citation>

diff --git a/tools/ucsc_tools/maftoaxt/.shed.yml b/tools/ucsc_tools/maftoaxt/.shed.yml
@@ -3,7 +3,7 @@ owner: iuc
 description: Convert dataset from MAF to axt format
 long_description: Convert MAF dataset to axt format. Depending on parameters and the number of alignments in the MAF file this can be a full conversion or data extraction.
 homepage_url: http://hgdownload.cse.ucsc.edu/admin/exe/
-remote_repository_url: https://github.com/galaxyproject/tools-iuc/tree/master/tools/ucsc-tools/maftoaxt
+remote_repository_url: https://github.com/galaxyproject/tools-iuc/tree/main/tools/ucsc_tools/maftoaxt
 type: unrestricted
 categories:
 - Convert Formats
diff --git a/tools/ucsc_tools/maftoaxt/maftoaxt.xml b/tools/ucsc_tools/maftoaxt/maftoaxt.xml
@@ -3,7 +3,7 @@
         Convert file from MAF to axt format
     </description>
     <macros>
-        <token name="@TOOL_VERSION@">377</token>
+        <token name="@TOOL_VERSION@">469</token>
     </macros>
     <xrefs>
         <xref type="bio.tools">UCSC_Genome_Browser_Utilities</xref>
@@ -25,7 +25,7 @@
             $stripDb
         ]]></command>
         <inputs>
-            <param name="in_maf" format="maf" type="data" label="maf file" />
+            <param name="in_maf" format="maf" type="data" label="MAF dataset" />
 
             <conditional name="tar_seq">
                 <param name="tarSeq" type="select" label="Select target sequence" optional="false" help="Which sequence from the MAF file should be used as the target sequence. ">
@@ -62,9 +62,14 @@
         <help><![CDATA[
     **What it does**
 
-    `mafToAxt`_ is a tool to convert maf dataset to axt format. Depending on parameters and the number of alignments in the MAF file this can be a full conversion or data extraction.
+    `mafToAxt`_ is a tool to convert `MAF`_ dataset to `axt`_ format. Depending on parameters and the number of alignments in the MAF file this can be a full conversion or data extraction.
+
+    For implementation details see mafToAxt's `source code`_.
 
     .. _mafToAxt: http://hgdownload.cse.ucsc.edu/admin/exe/linux.x86_64/FOOTER.txt
+    .. _axt: https://genome.ucsc.edu/goldenPath/help/axt.html
+    .. _MAF: https://genome.ucsc.edu/FAQ/FAQformat.html#format5
+    .. _source code: https://github.com/ucscGenomeBrowser/kent/blob/master/src/hg/mouseStuff/mafToAxt/mafToAxt.c
         ]]>    </help>
         <citations>
             <citation type="doi">10.1093/bib/bbs038</citation>

diff --git a/tools/ucsc_tools/twobittofa/.lint_skip b/tools/ucsc_tools/twobittofa/.lint_skip
diff --git a/tools/ucsc_tools/twobittofa/.shed.yml b/tools/ucsc_tools/twobittofa/.shed.yml
@@ -3,7 +3,7 @@ owner: iuc
 description: twoBitToFa is a tool to convert all or part of .2bit file to FASTA
 long_description: twoBitToFa is a tool to convert all or part of .2bit file to FASTA
 homepage_url: https://genome.ucsc.edu/goldenpath/help/twoBit.html
-remote_repository_url: https://github.com/galaxyproject/tools-iuc/tree/master/tools/ucsc-tools/twobittofa
+remote_repository_url: https://github.com/galaxyproject/tools-iuc/tree/main/tools/ucsc_tools/twobittofa
 type: unrestricted
 categories:
   - Convert Formats
diff --git a/tools/ucsc_tools/twobittofa/twobittofa.xml b/tools/ucsc_tools/twobittofa/twobittofa.xml
@@ -1,11 +1,11 @@
-<tool id="ucsc-twobittofa" name="twoBitToFa" version="@TOOL_VERSION@">
+<tool id="ucsc-twobittofa" name="twoBitToFa" version="@TOOL_VERSION@" profile="22.05">
     <description>Convert all or part of .2bit file to FASTA</description>
+    <macros>
+        <token name="@TOOL_VERSION@">469</token>
+    </macros>
     <xrefs>
         <xref type="bio.tools">UCSC_Genome_Browser_Utilities</xref>
     </xrefs>
-    <macros>
-        <token name="@TOOL_VERSION@">377</token>
-    </macros>
     <requirements>
         <requirement type="package" version="@TOOL_VERSION@">ucsc-twobittofa</requirement>
     </requirements>
@@ -57,10 +57,16 @@ $noMask
     <help><![CDATA[
 **What it does**
 
-twoBitToFa is a tool to convert all or part of .2bit file to FASTA.
+`twoBitToFa`_ is a tool to convert all or part of .2bit file to FASTA.
 
 For more information, check the `user manual <https://genome.ucsc.edu/goldenpath/help/twoBit.html>`_.
+
+For implementation details see twoBitToFa's `source code`_.
+
+.. _twoBitToFa: http://hgdownload.cse.ucsc.edu/admin/exe/linux.x86_64/FOOTER.txt
+.. _source code: https://github.com/ucscGenomeBrowser/kent/blob/master/src/utils/twoBitToFa/twoBitToFa.c
     ]]></help>
     <citations>
+        <citation type="doi">10.1093/bib/bbs038</citation>
     </citations>
 </tool>
diff --git a/tools/ucsc_tools/ucsc_axtchain/.shed.yml b/tools/ucsc_tools/ucsc_axtchain/.shed.yml
@@ -3,7 +3,7 @@ owner: iuc
 description: Chain together genome alignments
 long_description: Create a chain file from a genome alignment. The chain format describes a pairwise alignment that allow gaps in both sequences simultaneously.
 homepage_url: http://hgdownload.cse.ucsc.edu/admin/exe/
-remote_repository_url: https://github.com/galaxyproject/tools-iuc/tree/master/tools/ucsc-tools/axtchain
+remote_repository_url: https://github.com/galaxyproject/tools-iuc/tree/main/tools/ucsc_tools/axtchain
 type: unrestricted
 categories:
 - Sequence Analysis
diff --git a/tools/ucsc_tools/ucsc_axtchain/axtchain.xml b/tools/ucsc_tools/ucsc_axtchain/axtchain.xml
@@ -3,7 +3,7 @@
         Chain together axt alignments
     </description>
     <macros>
-        <token name="@TOOL_VERSION@">455</token>
+        <token name="@TOOL_VERSION@">469</token>
     </macros>
     <xrefs>
         <xref type="bio.tools">UCSC_Genome_Browser_Utilities</xref>
@@ -70,7 +70,7 @@
     <help><![CDATA[
 **What it does**
 
-`axtChain`_ is a tool to chain together alignments. The `chain`_ format describes a pairwise alignment that allow gaps in both sequences simultaneously.
+`axtChain`_ is a tool to chain together alignments in `axt`_ format. The `chain`_ format describes a pairwise alignment that allow gaps in both sequences simultaneously.
 
 
 sample linearGap file (loose)::
@@ -82,9 +82,12 @@ sample linearGap file (loose)::
     tGap    325     360     400     450     600     1100    3600    7600    15600   31600   56600
     bothGap 625     660     700     750     900     1400    4000    8000    16000   32000   57000
 
+For implementation details see axtChain's `source code`_.
 
 .. _axtChain: http://hgdownload.cse.ucsc.edu/admin/exe/linux.x86_64/FOOTER.txt
 .. _chain: https://genome.ucsc.edu/goldenPath/help/chain.html
+.. _axt: https://genome.ucsc.edu/goldenPath/help/axt.html
+.. _source code: https://github.com/ucscGenomeBrowser/kent/blob/master/src/hg/mouseStuff/axtChain/axtChain.c
     ]]>    </help>
     <citations>
         <citation type="doi">10.1093/bib/bbs038</citation>

diff --git a/tools/ucsc_tools/ucsc_axttomaf/.shed.yml b/tools/ucsc_tools/ucsc_axttomaf/.shed.yml
@@ -3,7 +3,7 @@ owner: iuc
 description: Convert dataset from axt to MAF format.
 long_description: This is a tool from the UCSC suite to convert dataset from axt to MAF format.
 homepage_url: http://hgdownload.cse.ucsc.edu/admin/exe/
-remote_repository_url: https://github.com/galaxyproject/tools-iuc/tree/master/tools/ucsc-tools/axttomaf
+remote_repository_url: https://github.com/galaxyproject/tools-iuc/tree/main/tools/ucsc_tools/axttomaf
 type: unrestricted
 categories:
 - Convert Formats
diff --git a/tools/ucsc_tools/ucsc_axttomaf/axttomaf.xml b/tools/ucsc_tools/ucsc_axttomaf/axttomaf.xml
@@ -3,7 +3,7 @@
         Convert dataset from axt to MAF format
     </description>
     <macros>
-        <token name="@TOOL_VERSION@">377</token>
+        <token name="@TOOL_VERSION@">469</token>
     </macros>
     <xrefs>
         <xref type="bio.tools">UCSC_Genome_Browser_Utilities</xref>
@@ -95,9 +95,12 @@
 
 `axtToMaf`_ is a tool to convert dataset from `axt format`_ to `MAF format`_.
 
+For implementation details see axtToMaf's `source code`_.
+
 .. _axtToMaf: http://hgdownload.cse.ucsc.edu/admin/exe/linux.x86_64/FOOTER.txt
 .. _axt format: https://genome.ucsc.edu/goldenPath/help/axt.html
 .. _MAF format: https://genome.ucsc.edu/FAQ/FAQformat.html#format5
+.. _source code: https://github.com/ucscGenomeBrowser/kent/blob/master/src/hg/mouseStuff/axtToMaf/axtToMaf.c
     ]]>    </help>
     <citations>
         <citation type="doi">10.1093/bib/bbs038</citation>

diff --git a/tools/ucsc_tools/ucsc_chainnet/.shed.yml b/tools/ucsc_tools/ucsc_chainnet/.shed.yml
@@ -3,7 +3,7 @@ owner: iuc
 description: Make alignment nets out of chains
 long_description:
 homepage_url: http://hgdownload.cse.ucsc.edu/admin/exe/
-remote_repository_url: https://github.com/galaxyproject/tools-iuc/tree/master/tools/ucsc-tools/ucsc_chainnet
+remote_repository_url: https://github.com/galaxyproject/tools-iuc/tree/main/tools/ucsc_tools/ucsc_chainnet
 type: unrestricted
 categories:
 - Sequence Analysis