fast5 to pod5 conversion

usegalaxy-au · Jun 17, 2024 · ce00a56 · ce00a56
1 parent 604d9d9
commit ce00a56
Show file tree

Hide file tree

Showing 5 changed files with 111 additions and 5 deletions.
diff --git a/tools/dorado/.shed.yml b/tools/dorado/.shed.yml
@@ -0,0 +1,23 @@
+---
+auto_tool_repositories:
+  name_template: "{{ tool_id }}"
+  description_template: "{{ tool_name }} from the dorado suite"
+categories:
+  - Sequence Analysis
+description: Dorado is a high-performance, easy-to-use, open source basecaller for Oxford Nanopore reads.
+exclude:
+  - tool_test_output.html
+  - tool_test_output.json
+homepage_url: https://github.com/nanoporetech/dorado
+long_description: >
+  Dorado is a high-performance, easy-to-use, open source basecaller for Oxford Nanopore reads.
+name: dorado
+owner: galaxy-australia
+remote_repository_url: https://github.com/usegalaxy-au/tools-au/tree/main/tools/dorado
+suite:
+  name: suite_dorado
+  description: >
+    Dorado is a high-performance, easy-to-use, open source basecaller for Oxford Nanopore reads.
+  long_description: >
+    Dorado is a high-performance, easy-to-use, open source basecaller for Oxford Nanopore reads.
+type: unrestricted
diff --git a/tools/dorado/dorado.xml b/tools/dorado/dorado.xml
@@ -25,8 +25,8 @@ calls.bam
 
     ]]></command>
     <inputs>
-        <param name="pod5_file" type="data" format="fast5" label="Raw fast5 file"/>
-        <param name="model" type="select" label="Basecalling model. See the Help section for info on model names." >
+        <param name="pod5_file" type="data" format="fast5" label="Raw pod5 file" help="Only pod5 is supported. You can convert fast5 to pod5 with the fast5_to_pod5 tool."/>
+        <param name="model" type="select" label="Basecalling model. See the Help section for info on model names.">
             <options from_data_table="dorado_models">
                 <!-- only allow models that shipped in this container -->
                 <filter type="static_value" column="1" value="@CONTAINER_HASH@"/>
@@ -40,6 +40,43 @@ calls.bam
         </param>
         <param type="select" argument="--kit-name" optional="true" label="Enable barcoding with the selected kit name." help="Reads are classified into their barcode groups during basecalling. The classification will be reflected in the read group name as well as in the BC tag of the output record.">
             <option value="EXP-NBD103">EXP-NBD103</option>
+            <option value="EXP-NBD104">EXP-NBD104</option>
+            <option value="EXP-NBD114">EXP-NBD114</option>
+            <option value="EXP-NBD196">EXP-NBD196</option>
+            <option value="EXP-PBC001">EXP-PBC001</option>
+            <option value="EXP-PBC096">EXP-PBC096</option>
+            <option value="SQK-16S024">SQK-16S024</option>
+            <option value="SQK-16S114-24">SQK-16S114-24</option>
+            <option value="SQK-LWB001">SQK-LWB001</option>
+            <option value="SQK-MLK111-96-XL">SQK-MLK111-96-XL</option>
+            <option value="SQK-MLK114-96-XL">SQK-MLK114-96-XL</option>
+            <option value="SQK-NBD111-24">SQK-NBD111-24</option>
+            <option value="SQK-NBD111-96">SQK-NBD111-96</option>
+            <option value="SQK-NBD114-24">SQK-NBD114-24</option>
+            <option value="SQK-NBD114-96">SQK-NBD114-96</option>
+            <option value="SQK-PBK004">SQK-PBK004</option>
+            <option value="SQK-PCB109">SQK-PCB109</option>
+            <option value="SQK-PCB110">SQK-PCB110</option>
+            <option value="SQK-PCB111-24">SQK-PCB111-24</option>
+            <option value="SQK-PCB114-24">SQK-PCB114-24</option>
+            <option value="SQK-RAB201">SQK-RAB201</option>
+            <option value="SQK-RAB204">SQK-RAB204</option>
+            <option value="SQK-RBK001">SQK-RBK001</option>
+            <option value="SQK-RBK004">SQK-RBK004</option>
+            <option value="SQK-RBK110-96">SQK-RBK110-96</option>
+            <option value="SQK-RBK111-24">SQK-RBK111-24</option>
+            <option value="SQK-RBK111-96">SQK-RBK111-96</option>
+            <option value="SQK-RBK114-24">SQK-RBK114-24</option>
+            <option value="SQK-RBK114-96">SQK-RBK114-96</option>
+            <option value="SQK-RLB001">SQK-RLB001</option>
+            <option value="SQK-RPB004">SQK-RPB004</option>
+            <option value="SQK-RPB114-24">SQK-RPB114-24</option>
+            <option value="TWIST-16-UDI">TWIST-16-UDI</option>
+            <option value="TWIST-96A-UDI">TWIST-96A-UDI</option>
+            <option value="VSK-PTC001">VSK-PTC001</option>
+            <option value="VSK-VMK001">VSK-VMK001</option>
+            <option value="VSK-VMK004">VSK-VMK004</option>
+            <option value="VSK-VPS001">VSK-VPS001</option>
         </param>
     </inputs>
     <outputs>
@@ -77,13 +114,30 @@ calls.bam
                 </assert_contents>
             </output>
         </test>
+        <!-- TODO: need a test for barcoded data. -->
+        <test expect_num_outputs="2">
+            <param name="pod5_file" value="FAL00375_473bf0ed_0.ten_reads.pod5"/>
+            <param name="model" value="dna_r9.4.1_e8_fast@v3.4"/>
+            <param name="trim" value="adapters"/>
+            <param name="kit_name" value="EXP-NBD103"/>
+            <output name="out_bam">
+                <assert_contents>
+                    <has_size value="60692" delta="6000"/>
+                </assert_contents>
+            </output>
+            <output name="out_tsv">
+                <assert_contents>
+                    <has_size value="1651" delta="200"/>
+                </assert_contents>
+            </output>
+        </test>
     </tests>
     <help><![CDATA[
 Basecall raw Nanopore data using Oxford Nanopore’s open source
 `dorado <https://github.com/nanoporetech/dorado/>`__ basecaller.
 
-The input is ``pod5`` format. If you have older data in fast5 format,
-you can convert them using the pod5 convert tool.
+The input is pod5 format. If you have older data in fast5 format, you
+can convert them using the ``fast5 to pod5`` convert tool.
 
 Basecalling models
 ------------------

diff --git a/tools/dorado/dorado_pod5_convert.xml b/tools/dorado/dorado_pod5_convert.xml
@@ -0,0 +1,29 @@
+<tool id="dorado_pod5_convert" name="fast5 to pod5" description="converter for raw Oxford Nanopore data" version="@VERSION@+galaxy0" python_template_version="3.5" profile="21.05">
+    <macros>
+        <import>macros.xml</import>
+    </macros>
+    <expand macro="xrefs"/>
+    <expand macro="requirements"/>
+    <command><![CDATA[
+    
+    ln -s '$fast5_in' ./input.fast5 &&
+    pod5 convert fast5 input.fast5 --output output.pod5
+
+    ]]>
+    </command>
+    <inputs>
+        <param name="fast5_in" type="data" format="fast5" label="Oxford Nanopore raw data in fast5 format"/>
+    </inputs>
+    <outputs>
+        <data format="fast5" name="pod5_out" label="Oxford Nanopore raw data in in pod5 format" from_work_dir="output.pod5"/>
+    </outputs>
+    <tests>
+        <test expect_num_outputs="1">
+            <param name="fast5_in" value="FAL00375_473bf0ed_0.ten_reads.0_0.fast5"/>
+            <output name="pod5_out" file="convert.test1.pod5"/>
+        </test>
+    </tests>
+    <help><![CDATA[
+        Convert fast5 to `pod5 <https://github.com/nanoporetech/pod5-file-format>`__ for basecalling with Dorado.
+            ]]></help>
+</tool>
diff --git a/tools/dorado/macros.xml b/tools/dorado/macros.xml
@@ -1,6 +1,6 @@
 <macros>
     <!-- UPDATING: pull the latest container and check the version. Update both tokens. You MUST also update the model list. See README.md for more.  -->
-    <token name="@VERSION@">0.7.1</token>
+    <token name="@VERSION@">0.7.1+80da5f5</token>
     <token name="@CONTAINER_HASH@">1c65eb070a9fc1d88710c4dc09b06541f96fdd28</token>
     <xml name="requirements">
         <requirements>

diff --git a/tools/dorado/test-data/FAL00375_473bf0ed_0.ten_reads.0_0.fast5 b/tools/dorado/test-data/FAL00375_473bf0ed_0.ten_reads.0_0.fast5