ARTbio · drosofff · Dec 15, 2024 · Oct 30, 2024 · Oct 30, 2024 · Oct 31, 2024
diff --git a/.github/workflows/pr.yaml b/.github/workflows/pr.yaml
@@ -14,7 +14,7 @@ on:
       - '*'
 env:
   GALAXY_FORK: galaxyproject
-  GALAXY_BRANCH: release_23.2
+  GALAXY_BRANCH: release_23.1
   MAX_CHUNKS: 4
   MAX_FILE_SIZE: 2M
 concurrency:

diff --git a/tools/wisecondorx/.shed.yml b/tools/wisecondorx/.shed.yml
@@ -0,0 +1,13 @@
+# .shed.yml supporting automatic pushes.
+owner: artbio
+name: wisecondorx
+description: WisecondorX
+long_description: |
+  Improved copy number detection for routine shallow whole-genome sequencing.
+  See homepage URL for manual and code.
+categories:
+  - Variant Analysis
+homepage_url: https://github.com/CenterForMedicalGeneticsGhent/WisecondorX/tree/master
+remote_repository_url: https://github.com/ARTbio/tools-artbio/tree/master/tools/wisecondorx
+toolshed:
+  - toolshed
diff --git a/tools/wisecondorx/macro.xml b/tools/wisecondorx/macro.xml
@@ -0,0 +1,54 @@
+<macros>
+    <token name="@VERSION@">1.2.9</token>
+    <token name="@WRAPPER_VERSION@">@VERSION@+galaxy0</token>
+    <token name="@PROFILE@">23.0</token>
+    <token name="@pipefail@"><![CDATA[set -o | grep -q pipefail && set -o pipefail;]]></token>
+
+    <xml name="requirements">
+        <requirements>
+            <requirement type="package" version="@VERSION@">wisecondorx</requirement>
+        </requirements>
+    </xml>
+<token name="@help@"><![CDATA[
+**What it does**
+
+WisecondorX, which uses a within-sample normalization technique, detects Copy
+Number Variation from BAM input files.
+
+It is important that **no** read quality filtering is executed prior to running
+WisecondorX: this software requires low-quality reads to distinguish informative
+bins from non-informative ones.
+
+There are three main stages (converting, reference build and predicting) when
+using WisecondorX:
+
+**1. Convert .bam files** of aligned reads to .npz files (for both normal and
+tumor samples) using the Galaxy tool **WisecondorX convert bam to npz**
+
+**2. Buid a reference index** from .npz files from **normal** samples using the
+Galaxy tool **WisecondorX build reference**.
+
+.. class:: warningmark
+
+Automated gender prediction, required to consistently analyze sex chromosomes,
+is based on a Gaussian mixture model. If few samples (<20) are included during
+reference creation, or not both male and female samples (for NIPT, this means
+male and female feti) are represented, this process might not be accurate.
+Therefore, alternatively, one can manually tweak the --yfrac parameter.
+
+.. class:: warningmark
+
+It is of paramount importance that the reference set consists of exclusively
+negative (normal) control samples that originate from the same sequencer, mapper,
+reference genome, type of material, ... etc, as the test samples. As a rule of
+thumb, think of all laboratory and in silico steps: the more sources of bias that
+can be omitted, the better.
+
+Try to include at least 50 samples per reference. The more the better, yet, from
+500 on it is unlikely to observe additional improvement concerning normalization.
+
+**3. Predict Copy Number Variantions** from the reference index and tumor .npz cases
+of interest using the Galaxy tool **WisecondorX predict CNVs**
+
+]]></token>
+</macros>
diff --git a/tools/wisecondorx/test-data/0.ref.npz b/tools/wisecondorx/test-data/0.ref.npz
diff --git a/tools/wisecondorx/test-data/1.ref.npz b/tools/wisecondorx/test-data/1.ref.npz
diff --git a/tools/wisecondorx/test-data/2.ref.npz b/tools/wisecondorx/test-data/2.ref.npz
diff --git a/tools/wisecondorx/test-data/3.ref.npz b/tools/wisecondorx/test-data/3.ref.npz
diff --git a/tools/wisecondorx/test-data/4.ref.npz b/tools/wisecondorx/test-data/4.ref.npz
diff --git a/tools/wisecondorx/test-data/5.ref.npz b/tools/wisecondorx/test-data/5.ref.npz
diff --git a/tools/wisecondorx/test-data/6.ref.npz b/tools/wisecondorx/test-data/6.ref.npz
diff --git a/tools/wisecondorx/test-data/7.ref.npz b/tools/wisecondorx/test-data/7.ref.npz
diff --git a/tools/wisecondorx/test-data/8.ref.npz b/tools/wisecondorx/test-data/8.ref.npz
diff --git a/tools/wisecondorx/test-data/9.ref.npz b/tools/wisecondorx/test-data/9.ref.npz
diff --git a/tools/wisecondorx/test-data/chr1.png b/tools/wisecondorx/test-data/chr1.png
diff --git a/tools/wisecondorx/test-data/chr10.png b/tools/wisecondorx/test-data/chr10.png
diff --git a/tools/wisecondorx/test-data/chr11.png b/tools/wisecondorx/test-data/chr11.png
diff --git a/tools/wisecondorx/test-data/chr12.png b/tools/wisecondorx/test-data/chr12.png
diff --git a/tools/wisecondorx/test-data/chr13.png b/tools/wisecondorx/test-data/chr13.png
diff --git a/tools/wisecondorx/test-data/chr14.png b/tools/wisecondorx/test-data/chr14.png
diff --git a/tools/wisecondorx/test-data/chr15.png b/tools/wisecondorx/test-data/chr15.png
diff --git a/tools/wisecondorx/test-data/chr16.png b/tools/wisecondorx/test-data/chr16.png
diff --git a/tools/wisecondorx/test-data/chr17.png b/tools/wisecondorx/test-data/chr17.png
diff --git a/tools/wisecondorx/test-data/chr18.png b/tools/wisecondorx/test-data/chr18.png
diff --git a/tools/wisecondorx/test-data/chr19.png b/tools/wisecondorx/test-data/chr19.png
diff --git a/tools/wisecondorx/test-data/chr2.png b/tools/wisecondorx/test-data/chr2.png
diff --git a/tools/wisecondorx/test-data/chr20.png b/tools/wisecondorx/test-data/chr20.png
diff --git a/tools/wisecondorx/test-data/chr21.png b/tools/wisecondorx/test-data/chr21.png
diff --git a/tools/wisecondorx/test-data/chr22.png b/tools/wisecondorx/test-data/chr22.png
diff --git a/tools/wisecondorx/test-data/chr3.png b/tools/wisecondorx/test-data/chr3.png
diff --git a/tools/wisecondorx/test-data/chr4.png b/tools/wisecondorx/test-data/chr4.png
diff --git a/tools/wisecondorx/test-data/chr5.png b/tools/wisecondorx/test-data/chr5.png
diff --git a/tools/wisecondorx/test-data/chr6.png b/tools/wisecondorx/test-data/chr6.png
diff --git a/tools/wisecondorx/test-data/chr7.png b/tools/wisecondorx/test-data/chr7.png
diff --git a/tools/wisecondorx/test-data/chr8.png b/tools/wisecondorx/test-data/chr8.png
diff --git a/tools/wisecondorx/test-data/chr9.png b/tools/wisecondorx/test-data/chr9.png
diff --git a/tools/wisecondorx/test-data/chrX.png b/tools/wisecondorx/test-data/chrX.png
diff --git a/tools/wisecondorx/test-data/genome_wide.png b/tools/wisecondorx/test-data/genome_wide.png
diff --git a/tools/wisecondorx/test-data/input_reference_predict.10000kb.npz b/tools/wisecondorx/test-data/input_reference_predict.10000kb.npz
diff --git a/tools/wisecondorx/test-data/input_sample_predict.npz b/tools/wisecondorx/test-data/input_sample_predict.npz
diff --git a/tools/wisecondorx/test-data/npz_convert_input.bam b/tools/wisecondorx/test-data/npz_convert_input.bam
diff --git a/tools/wisecondorx/test-data/npz_convert_output.npz b/tools/wisecondorx/test-data/npz_convert_output.npz
diff --git a/tools/wisecondorx/test-data/output_reference.npz b/tools/wisecondorx/test-data/output_reference.npz
diff --git a/tools/wisecondorx/test-data/predict_abberations.bed b/tools/wisecondorx/test-data/predict_abberations.bed
@@ -0,0 +1 @@
+chr	start	end	ratio	zscore	type
diff --git a/tools/wisecondorx/test-data/predict_bins.bed b/tools/wisecondorx/test-data/predict_bins.bed
diff --git a/tools/wisecondorx/test-data/predict_segments.bed b/tools/wisecondorx/test-data/predict_segments.bed
@@ -0,0 +1,28 @@
+chr	start	end	ratio	zscore
+1	1	120000000	-0.0087	-2.538405323843277
+1	140000001	250000000	0.0176	4.312862224344486
+2	1	250000000	0.0077	3.3452597487045215
+3	1	200000000	0.0058	2.311500455599722
+4	1	40000000	0.0007	0.11656739335236131
+4	40000001	190000000	0.0006	0.35088837729023276
+5	1	190000000	-0.0138	-4.742762233161156
+6	1	170000000	0.0029	0.9290111777306987
+7	1	160000000	-0.0066	-1.3173166420256102
+8	1	150000000	-0.0093	-2.345755449706625
+9	1	20000000	0.0057	0.5378706474020528
+9	20000001	40000000	-0.0068	-0.5822999425114906
+9	60000001	140000000	-0.0066	-0.8909887079018787
+10	1	140000000	-0.0053	-0.972447941393507
+11	1	140000000	-0.0011	-0.25869752816517777
+12	1	140000000	-0.0067	-1.6120716034316234
+13	20000001	120000000	0.0091	2.6165198496794595
+14	20000001	110000000	0.0046	1.4493631003182375
+15	30000001	110000000	0.0022	0.5631104559578264
+16	1	90000000	0.0009	0.20349127260764568
+17	1	90000000	0.0198	3.2916417620408067
+18	1	90000000	0.0209	2.8871676559099937
+19	10000001	60000000	-0.0008	-0.09701491554952066
+20	1	70000000	0.0034	0.49591449488386785
+21	10000001	50000000	-0.0053	-0.6285903890420564
+22	10000001	60000000	-0.0177	-1.875889555547745
+X	1	160000000	-0.0134	-2.6093805092022024
diff --git a/tools/wisecondorx/test-data/predict_statistics.txt b/tools/wisecondorx/test-data/predict_statistics.txt
@@ -0,0 +1,29 @@
+chr	ratio.mean	ratio.median	zscore
+1	0.003256724685060297	0.009032538940713365	1.1604576995182534
+2	0.007728111371764734	0.00879518825265517	3.6168182974192224
+3	0.0057970691646083835	0.007918358867360545	2.2102487208570447
+4	0.0005794303608920617	-0.0038693714215764057	0.3050780556626696
+5	-0.013750074657332427	-0.00306199917667051	-5.00766851062217
+6	0.00293681601872286	-0.0016253998352964258	0.9407744372804929
+7	-0.0066155513952186815	-0.007596035725320216	-1.3291500169147508
+8	-0.009253570570903735	-0.01001300559828008	-2.3586673175031008
+9	-0.004174118616434959	-0.0024356973032108246	-0.7243064497148664
+10	-0.005266146751525984	0.0028177250094220192	-1.0963168057162755
+11	-0.0010682214397647377	-0.012687471128064745	-0.25403103304752794
+12	-0.006720586939014568	-0.007785953193232591	-1.7313975674091997
+13	0.009139617144346813	0.002351198145606662	2.7174009747784944
+14	0.004617644740226149	0.0035058727944338825	1.4966978929197126
+15	0.0021565565774295426	-0.005083704704282363	0.5800013406347537
+16	0.0009244703405124733	0.007301717962691849	0.20870556732386122
+17	0.01982791519920036	0.006869817773389827	3.1146698354398854
+18	0.020898589066626748	0.009665993221586318	2.845910113450034
+19	-0.000657740980105667	0.0016192329112033454	-0.08006761423301532
+20	0.003446546608787667	0.002058925822068644	0.5033887759993748
+21	-0.005290652555760193	-0.011028957124213846	-0.5341417465937656
+22	-0.017713587100928845	-0.0327155577009055	-1.6790692284983704
+X	-0.013378674115274611	-0.015072799437654107	-2.6107602961602137
+Gender based on --yfrac (or manually overridden by --gender): F
+Number of reads: 2081160
+Standard deviation of the ratios per chromosome: 0.00932
+Median segment variance per bin (doi: 10.1093/nar/gky1263): 0.0008
+Copy number profile abnormality (CPA) score (doi: 10.1186/s13073-020-00735-4): 2.32398
diff --git a/tools/wisecondorx/wisecondor_npz_converter.xml b/tools/wisecondorx/wisecondor_npz_converter.xml
@@ -0,0 +1,57 @@
+<tool id="wisecondorx_npz_converter" name="WisecondorX convert" version="@WRAPPER_VERSION@" profile="@PROFILE@">
+    <description>
+        bam to npz
+    </description>
+    <macros>
+        <import>macro.xml</import>
+    </macros>
+    <expand macro="requirements"/>
+    <stdio>
+        <exit_code range="1:" level="fatal" description="Error occured" />
+    </stdio>
+    <command detect_errors="exit_code"><![CDATA[
+    @pipefail@
+    ln -f -s $bam.metadata.bam_index input.bam.bai &&
+    ln -f -s $bam input.bam &&
+    printf "Creating 5kb bins for file $bam.element_identifier" &&
+    WisecondorX convert input.bam output.npz
+    ]]></command>
+    <inputs>
+        <param name="bam" type="data" label="Bam input" format="bam"
+               help="input Bam is converted in .npz file"/>
+     </inputs>
+    <outputs>
+        <data name="npz" format="npz" from_work_dir="output.npz" label="${on_string}.npz" />
+    </outputs>
+    <tests>
+        <test expect_num_outputs="1">
+            <param ftype="bam" name="bam" value="npz_convert_input.bam" />
+            <output name="npz" ftype="npz" file="npz_convert_output.npz" compare="sim_size" delta="10000"/>
+        </test>
+    </tests>
+    <help>
+@help@
+<![CDATA[
+.. class:: infomark
+
+**WisecondorX convert input.bam/cram output.npz [--optional arguments]**
+
+Option List::
+
+  --reference   Fasta reference to be used with cram inputs.
+                This option is currently not available in this Galaxy wrapper,
+                which takes only bam inputs.
+  --binsize     Size per bin in bp; the reference bin size should be a multiple of this value.
+                Note that this parameter does not impact the resolution, yet it
+                can be used to optimize processing speed (default: x=5e3).
+                The --binsize parameter is currently not exposed in this Galaxy
+                wrapper and is fixed to 5e3
+  --normdup     Use this flag to avoid duplicate removal.
+                The --normdup parameter is currently not exposed in this Galaxy
+                wrapper. Default is to remove duplicates.
+
+    ]]></help>
+    <citations>
+        <citation type="doi">10.1093/nar/gky1263</citation>
+    </citations>
+</tool>
diff --git a/tools/wisecondorx/wisecondor_predict.xml b/tools/wisecondorx/wisecondor_predict.xml
@@ -0,0 +1,112 @@
+<tool id="wisecondorx_predict" name="WisecondorX predict" version="@WRAPPER_VERSION@" profile="@PROFILE@">
+    <description>
+        CNVs
+    </description>
+    <macros>
+        <import>macro.xml</import>
+    </macros>
+    <expand macro="requirements"/>
+    <stdio>
+        <exit_code range="1:" level="fatal" description="Error occured" />
+    </stdio>
+    <command detect_errors="exit_code"><![CDATA[
+    @pipefail@
+    ln -s $npz_input sample.npz &&
+    ln -s $reference reference.npz &&
+    WisecondorX predict sample.npz reference.npz sample --plot --bed
+    ]]></command>
+    <inputs>
+        <param name="npz_input" type="data" format="npz" label="npz file from sample to analyse"/>
+        <param name="reference" type="data" format="npz" label="npz reference built with WisecondorX build"/>
+    </inputs>
+    <outputs>
+        <data name="aberrations" format="bed" label="sample abberations" from_work_dir="sample_aberrations.bed"/>
+        <data name="bins" format="bed" label="sample bins" from_work_dir="sample_bins.bed"/>
+        <data name="segments" format="bed" label="sample segments" from_work_dir="sample_segments.bed"/>
+        <data name="statistics" format="txt" label="statistics" from_work_dir="sample_statistics.txt"/>
+        <collection name="plots" type="list" format="png" label="CNV plots">
+            <discover_datasets pattern="__name_and_ext__" directory="sample.plots" />
+        </collection>
+    </outputs>
+    <tests>
+        <test expect_num_outputs="5">
+            <param name="npz_input" value="input_sample_predict.npz" ftype="npz" />
+            <param name="reference" value="input_reference_predict.10000kb.npz" />
+            <output name="aberrations" ftype="bed" file="predict_abberations.bed" />
+            <output name="bins" ftype="bed" file="predict_bins.bed" compare="sim_size" delta="1000"/>
+            <output name="segments" ftype="bed" file="predict_segments.bed"/>
+            <output name="statistics" ftype="txt" file="predict_statistics.txt" compare="sim_size" delta="1000"/>
+            <output_collection name="plots" type="list">
+                <element name="chr1" file="chr1.png" compare="sim_size" delta="10000"/>
+                <element name="chr10" file="chr10.png" compare="sim_size" delta="10000"/>
+                <element name="chr11" file="chr11.png" compare="sim_size" delta="10000"/>
+                <element name="chr12" file="chr12.png" compare="sim_size" delta="10000"/>
+                <element name="chr13" file="chr13.png" compare="sim_size" delta="10000"/>
+                <element name="chr14" file="chr14.png" compare="sim_size" delta="10000"/>
+                <element name="chr15" file="chr15.png" compare="sim_size" delta="10000"/>
+                <element name="chr16" file="chr16.png" compare="sim_size" delta="10000"/>
+                <element name="chr17" file="chr17.png" compare="sim_size" delta="10000"/>
+                <element name="chr18" file="chr18.png" compare="sim_size" delta="10000"/>
+                <element name="chr19" file="chr19.png" compare="sim_size" delta="10000"/>
+                <element name="chr2" file="chr2.png" compare="sim_size" delta="10000"/>
+                <element name="chr20" file="chr20.png" compare="sim_size" delta="10000"/>
+                <element name="chr21" file="chr21.png" compare="sim_size" delta="10000"/>
+                <element name="chr22" file="chr22.png" compare="sim_size" delta="10000"/>
+                <element name="chr3" file="chr3.png" compare="sim_size" delta="10000"/>
+                <element name="chr4" file="chr4.png" compare="sim_size" delta="10000"/>
+                <element name="chr5" file="chr5.png" compare="sim_size" delta="10000"/>
+                <element name="chr6" file="chr6.png" compare="sim_size" delta="10000"/>
+                <element name="chr7" file="chr7.png" compare="sim_size" delta="10000"/>
+                <element name="chr8" file="chr8.png" compare="sim_size" delta="10000"/>
+                <element name="chr9" file="chr9.png" compare="sim_size" delta="10000"/>
+                <element name="chrX" file="chrX.png" compare="sim_size" delta="10000"/>
+                <element name="genome_wide" file="genome_wide.png" compare="sim_size" delta="10000"/>
+            </output_collection>
+        </test>
+    </tests>
+    <help>
+@help@
+<![CDATA[
+.. class:: infomark
+
+**WisecondorX predict test_input.npz reference_input.npz output_id [--optional arguments]**
+
+Option List::
+
+  --minrefbins      Minimum amount of sensible reference bins per target bin;
+                    should generally not be tweaked (default: x=150)
+  --maskrepeats     Bins with distances > mean + sd * 3 in the reference will be
+                    masked. This parameter represents the number of masking cycles
+                    and defines the stringency of the blacklist (default: x=5)
+                    **Should be a multiple of the 5e3**.
+                    Not exposed in this Galaxy wrapper.
+  --zscore          Z-score cutoff to call segments as aberrations (default: x=5)
+  --alpha           P-value cutoff for calling circular binary segmentation
+                    breakpoints (default: x=1e-4).
+                    Not exposed in this Galaxy wrapper.
+  --beta            When beta is given, --zscore is ignored. Beta sets a ratio
+                    cutoff for aberration calling. It's a number between 0 (liberal)
+                    and 1 (conservative) and, when used, is optimally close to the
+                    purity (e.g. fetal/tumor fraction)
+                    Not exposed in this Galaxy wrapper.
+  --blacklist       Blacklist for masking additional regions; requires headerless
+                    .bed file. This is particularly useful when the reference set
+                    is too small to recognize some obvious loci (such as centromeres).
+                    Not exposed in this Galaxy wrapper.
+  --gender          Force WisecondorX to analyze this case as male (M) or female (F).
+                    Useful when e.g. dealing with a loss of chromosome Y, which
+                    causes erroneous gender predictions (choices: x=F or x=M).
+                    Not exposed in this Galaxy wrapper.
+  --bed             Outputs tab-delimited .bed files.
+  --plot            Outputs custom .png plots, directly interpretable.
+  --ylim  [a,b]     Force WisecondorX to use y-axis interval [a,b] during plotting, e.g. [-2,2].
+                    Not exposed in this Galaxy wrapper.
+  --cairo           Some operating systems require the cairo bitmap type to write plots.
+                    Not exposed in this Galaxy wrapper.
+  --seed            Random seed for segmentation algorithm (default:None).
+                    Not exposed in this Galaxy wrapper.
+    ]]></help>
+    <citations>
+        <citation type="doi">10.1093/nar/gky1263</citation>
+    </citations>
+</tool>
diff --git a/tools/wisecondorx/wisecondor_reference_builder.xml b/tools/wisecondorx/wisecondor_reference_builder.xml
@@ -0,0 +1,66 @@
+<tool id="wisecondorx_reference_builder" name="WisecondorX build" version="@WRAPPER_VERSION@" profile="@PROFILE@">
+    <description>
+        reference
+    </description>
+    <macros>
+        <import>macro.xml</import>
+    </macros>
+    <expand macro="requirements"/>
+    <stdio>
+        <exit_code range="1:" level="fatal" description="Error occured" />
+    </stdio>
+    <command detect_errors="exit_code"><![CDATA[
+    @pipefail@
+    #for $num, $file in enumerate($npz_inputs):
+        ln -s $file "${num}.npz" &&
+    #end for
+    WisecondorX newref *.npz reference.npz
+        --binsize ${bin}
+        --cpus \${GALAXY_SLOTS:-4} &&
+    mv reference.npz $npz
+    ]]></command>
+    <inputs>
+        <param name="npz_inputs" type="data" label="npz inputs" multiple="True" format="npz"
+               help="Build reference from npz inputs from normal sample (at least 10 samples required)"/>
+        <param name="bin" size="9" type="integer" value="100000" label="Bin size in nucleotides"
+               help="Bin default value is 100 kb (100000)" />
+     </inputs>
+    <outputs>
+        <data name="npz" format="npz" label="reference_${bin}nt" />
+    </outputs>
+    <tests>
+        <test expect_num_outputs="1">
+            <param name="npz_inputs"
+                   value="0.ref.npz,1.ref.npz,2.ref.npz,3.ref.npz,4.ref.npz,5.ref.npz,6.ref.npz,7.ref.npz,8.ref.npz,9.ref.npz"/>
+            <param name="bin" value="10000" />
+            <output name="npz" ftype="npz" file="output_reference.npz" compare="sim_size" delta="10000"/>
+        </test>
+    </tests>
+    <help>
+@help@
+<![CDATA[
+.. class:: infomark
+
+**WisecondorX newref reference_input_dir/*.npz reference_output.npz [--optional arguments]**
+
+Option List::
+
+  --nipt        Always include this flag for the generation of a NIPT reference
+  --binsize     Size per bin in bp, defines the resolution of the output (default: x=1e5)
+                **Should be a multiple of the 5e3**
+  --refsize     Amount of reference locations per target;
+                should generally not be tweaked (default: x=300)
+  --yfrac       Y read fraction cutoff, in order to manually define gender.
+                Setting this to 1 will treat all samples as female.
+                This parameter is not currently exposed in the Galaxy wrapper.
+  --plotyfrac   plots Y read fraction histogram and Gaussian mixture fit to file x,
+                can help when setting --yfrac manually; software quits after plotting
+                The --normdup parameter is currently not exposed in this Galaxy
+                wrapper. Default is to remove duplicates.
+  --cpus        Number of threads requested (This is defined by the Galaxy administrator)
+
+    ]]></help>
+    <citations>
+        <citation type="doi">10.1093/nar/gky1263</citation>
+    </citations>
+</tool>