Mercurial > repos > ebi-gxa > anndata_ops

--- a/anndata_operations.xml	Fri Apr 14 13:12:01 2023 +0000
+++ b/anndata_operations.xml	Sat Jul 15 08:44:46 2023 +0000
@@ -1,5 +1,5 @@
 <?xml version="1.0" encoding="utf-8"?>
-<tool id="anndata_ops" name="AnnData Operations" version="@TOOL_VERSION@+galaxy91" profile="@PROFILE@">
+<tool id="anndata_ops" name="AnnData Operations" version="@TOOL_VERSION@+galaxy92" profile="@PROFILE@">
   <description>modifies metadata and flags genes</description>
   <macros>
     <import>scanpy_macros2.xml</import>
@@ -50,6 +50,7 @@
   <configfiles>
     <configfile name="operations">
 import gc
+from os import makedirs
 import scanpy as sc
 import anndata
 from numpy import all
@@ -61,9 +62,9 @@
   appendents = (suffix + df.groupby(field).cumcount().astype(str).replace('0','')).replace(suffix, '')
   df[new_field] = df[field].astype(str) + appendents.astype(str)
   return df
-
+
 adata = sc.read('input.h5')
-
+
 #if $add_cell_metadata.default:
 import pandas as pd

@@ -96,7 +97,7 @@
               print(f"Changing {col} from {merged_obs[col].dtype} to {prev_dtype}")
               merged_obs[col] = merged_obs[col].astype(prev_dtype)
   return merged_obs
-
+
 adata.obs = add_cell_metadata(adata)
 #end if

@@ -104,6 +105,14 @@
 adata.raw = adata
 #end if

+#if $swap_layer_to_x.default:
+#if $swap_layer_to_x.new_name_x:
+adata.layers['${swap_layer_to_x.new_name_x}'] = adata.X
+#end if
+adata.X = adata.layers['${swap_layer_to_x.layer}']
+del adata.layers['${swap_layer_to_x.layer}']
+#end if
+
 gene_name = '${gene_symbols_field}'
 qc_vars = list()

@@ -286,13 +295,37 @@
 if 'n_counts' not in adata.var.columns:
     sc.pp.filter_genes(adata, min_counts=0)

+#if not $split_on_obs.default or $split_on_obs.output_main:
 adata.write('output.h5', compression='gzip')
+#end if
+
+#if $split_on_obs.default:
+s = 0
+res_dir = "output_split"
+makedirs(res_dir, exist_ok=True)
+for field_value in adata.obs["${split_on_obs.key}"].unique():
+    ad_s = adata[adata.obs.${split_on_obs.key} == field_value]
+    ad_s.write(f"{res_dir}/${split_on_obs.key}_{s}.h5", compression='gzip')
+    if s > 0:
+        gc.collect()
+    s += 1
+#end if
+
+
     </configfile>
 </configfiles>

   <inputs>
     <param name="input_obj_file" argument="input-object-file" type="data" format="h5,h5ad" label="Input object in hdf5 AnnData format"/>
     <expand macro="output_object_params_no_loom"/>
+    <conditional name="swap_layer_to_x">
+      <param name="default" type="boolean" checked="false" label="Swap layer to X"/>
+      <when value="true">
+        <param name="layer" type="text" value="" label="Name of layer to swap to X" help="This layer name needs to exist within ad.layers or this will fail."/>
+        <param name="new_name_x" type="text" value="old_X" label="Name of the new slot for X within layers" help="Leave empty and the old X will be lost."/>
+      </when>
+      <when value="false"/>
+    </conditional>
     <conditional name="add_cell_metadata">
       <param name="default" type="boolean" checked="false" label="Merge additional cell metadata"/>
       <when value="true">
@@ -387,10 +420,27 @@
       </when>
     </conditional>
     <param name="sanitize_varm" type="boolean" checked="false" label="Sanitise any null raw.varm objects if any" help="This might be relevant for interfacing with newer versions of AnnData, that might complain if .raw includes a varm null object."/>
+    <conditional name="split_on_obs">
+      <param name="default" type="boolean" checked="false" label="Split on obs" help="Split the AnnData object into multiple AnnData objects based on the values of a given obs key. This is useful for example to split a dataset based on a cluster annotation."/>
+      <when value="true">
+        <param name="key" type="text" label="Obs key to split on" help="The obs key to split on. For example, if you want to split on cluster annotation, you can use the key 'louvain'."/>
+        <param name="output_main" type="boolean" checked="true" label="Output main AnnData object" help="If checked, the main AnnData object will be outputted as well."/>
+      </when>
+      <when value="false"/>
+    </conditional>
   </inputs>

   <outputs>
-    <expand macro="output_data_obj_no_loom" description="metadata changes on"/>
+    <data name="output_h5ad" format="h5ad" from_work_dir="output.h5" label="${tool.name} on ${on_string}: @DESCRIPTION@ AnnData">
+      <filter>output_format == 'anndata_h5ad' and (('output_main' in split_on_obs and split_on_obs['output_main']) or (not split_on_obs['default']))</filter>
+    </data>
+    <data name="output_h5" format="h5" from_work_dir="output.h5" label="${tool.name} on ${on_string}: @DESCRIPTION@ AnnData (h5)">
+      <filter>output_format == 'anndata' and (('output_main' in split_on_obs and split_on_obs['output_main']) or (not split_on_obs['default']))</filter>
+    </data>
+    <collection name="output_h5ad_split" type="list" label="${tool.name} on ${on_string}: @DESCRIPTION@ AnnData split">
+      <discover_datasets pattern="(?P&lt;designation&gt;.+)\.h5" directory="output_split" format="h5ad" visible="true"/>
+      <filter>split_on_obs['default']</filter>
+    </collection>
   </outputs>

   <tests>
@@ -399,6 +449,19 @@
       <output name="output_h5ad" file="anndata_ops.h5" ftype="h5ad" compare="sim_size"/>
     </test>
     <test>
+      <param name="input_obj_file" value="mnn.h5"/>
+      <conditional name="swap_layer_to_x">
+        <param name="default" value="true"/>
+        <param name="layer" value="mnn"/>
+        <param name="new_name_x" value="X_old"/>
+      </conditional>
+      <output name="output_h5ad" ftype="h5ad">
+        <assert_contents>
+          <has_h5_keys keys="layers/X_old" />
+        </assert_contents>
+      </output>
+    </test>
+    <test>
       <param name="input_obj_file" value="anndata_ops.h5"/>
       <param name="from_var" value = "gene_symbols" />
       <param name="to_var" value = "hello_all" />
@@ -484,6 +547,26 @@
         </assert_contents>
       </output>
     </test>
+    <test>
+      <param name="input_obj_file" value="find_cluster.h5"/>
+      <conditional name="split_on_obs">
+        <param name="default" value="true"/>
+        <param name="key" value="louvain"/>
+        <param name="output_main" value="true"/>
+      </conditional>
+      <output name="output_h5ad" ftype="h5ad">
+        <assert_contents>
+          <has_h5_keys keys="obs/louvain" />
+        </assert_contents>
+      </output>
+      <output_collection name="output_h5ad_split" type="list" count="5">
+        <element name="louvain_0" ftype="h5ad">
+          <assert_contents>
+            <has_h5_keys keys="obs/louvain" />
+          </assert_contents>
+        </element>
+      </output_collection>
+    </test>
   </tests>

   <help><![CDATA[