Skip to content

Commit c204e61

Browse files
bbimberbbimber
committed
Allow cellbender to run against loupe and cite-seq files
1 parent 0a4b557 commit c204e61

File tree

3 files changed

+305
-0
lines changed

3 files changed

+305
-0
lines changed

‎singlecell/src/org/labkey/singlecell/SingleCellModule.java‎

Lines changed: 4 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -37,6 +37,8 @@
3737
import org.labkey.singlecell.analysis.ProcessSingleCellHandler;
3838
import org.labkey.singlecell.button.FeatureBarcodeButton;
3939
import org.labkey.singlecell.pipeline.singlecell.*;
40+
import org.labkey.singlecell.run.CellBenderCiteSeqHandler;
41+
import org.labkey.singlecell.run.CellBenderLoupeHandler;
4042
import org.labkey.singlecell.run.CellRangerFeatureBarcodeHandler;
4143
import org.labkey.singlecell.run.CellRangerGexCountStep;
4244
import org.labkey.singlecell.run.CellRangerVDJWrapper;
@@ -136,6 +138,8 @@ public static void registerPipelineSteps()
136138
SequenceAnalysisService.get().registerFileHandler(new CellRangerRawDataHandler());
137139
SequenceAnalysisService.get().registerFileHandler(new ProcessSingleCellHandler());
138140
SequenceAnalysisService.get().registerFileHandler(new ProcessSeuratObjectHandler());
141+
SequenceAnalysisService.get().registerFileHandler(new CellBenderCiteSeqHandler());
142+
SequenceAnalysisService.get().registerFileHandler(new CellBenderLoupeHandler());
139143

140144
//Single-cell:
141145
SequencePipelineService.get().registerPipelineStep(new AppendCiteSeq.Provider());

‎singlecell/src/org/labkey/singlecell/run/CellBenderCiteSeqHandler.java‎

Lines changed: 166 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,166 @@
1+
package org.labkey.singlecell.run;
2+
3+
import org.apache.commons.io.FileUtils;
4+
import org.json.JSONObject;
5+
import org.labkey.api.module.ModuleLoader;
6+
import org.labkey.api.pipeline.PipelineJob;
7+
import org.labkey.api.pipeline.PipelineJobException;
8+
import org.labkey.api.pipeline.RecordedAction;
9+
import org.labkey.api.sequenceanalysis.SequenceOutputFile;
10+
import org.labkey.api.sequenceanalysis.pipeline.AbstractParameterizedOutputHandler;
11+
import org.labkey.api.sequenceanalysis.pipeline.CommandLineParam;
12+
import org.labkey.api.sequenceanalysis.pipeline.SequenceAnalysisJobSupport;
13+
import org.labkey.api.sequenceanalysis.pipeline.SequenceOutputHandler;
14+
import org.labkey.api.sequenceanalysis.pipeline.SequencePipelineService;
15+
import org.labkey.api.sequenceanalysis.pipeline.ToolParameterDescriptor;
16+
import org.labkey.api.sequenceanalysis.run.SimpleScriptWrapper;
17+
import org.labkey.api.util.FileUtil;
18+
import org.labkey.singlecell.SingleCellModule;
19+
20+
import java.io.File;
21+
import java.io.IOException;
22+
import java.util.ArrayList;
23+
import java.util.Arrays;
24+
import java.util.List;
25+
26+
public class CellBenderCiteSeqHandler extends AbstractParameterizedOutputHandler<SequenceOutputHandler.SequenceOutputProcessor>
27+
{
28+
public CellBenderCiteSeqHandler()
29+
{
30+
super(ModuleLoader.getInstance().getModule(SingleCellModule.class), "Run CellBender (CITE-seq)", "This will run cellbender on the input cellranger folder and create a subset matrix with background/ambient noise removed.", null, getParams());
31+
}
32+
33+
protected static List<ToolParameterDescriptor> getParams()
34+
{
35+
return Arrays.asList(
36+
ToolParameterDescriptor.createCommandLineParam(CommandLineParam.create("--expected-cells"), "expectedCells", "Expected Cells", "Passed to CellBender --expected-cells", "ldk-integerfield", null, 5000),
37+
ToolParameterDescriptor.createCommandLineParam(CommandLineParam.create("--total-droplets-included"), "totalDropletsIncluded", "Total Droplets Included", "Passed to CellBender --total-droplets-included", "ldk-integerfield", null, 20000),
38+
ToolParameterDescriptor.createCommandLineParam(CommandLineParam.create("--fpr"), "fpr", "fpr Cells", "Passed to CellBender --fpr", "ldk-numberfield", new JSONObject(){{
39+
put("decimalPrecision", 3);
40+
}}, 0.01),
41+
ToolParameterDescriptor.createCommandLineParam(CommandLineParam.create("--epochs"), "epochs", "Epochs", "Passed to CellBender --epochs", "ldk-integerfield", null, 150),
42+
ToolParameterDescriptor.createCommandLineParam(CommandLineParam.createSwitch("--cuda"), "useGpus", "Use GPUs", "If checked, the --cuda argument will be set on cellbender", "checkbox", null, false)
43+
);
44+
}
45+
46+
@Override
47+
public boolean doSplitJobs()
48+
{
49+
return true;
50+
}
51+
52+
@Override
53+
public boolean canProcess(SequenceOutputFile o)
54+
{
55+
return "CITE-seq Counts".equals(o.getCategory()) & "matrix.mtx.gz".equals(o.getFile().getName());
56+
}
57+
58+
@Override
59+
public boolean doRunRemote()
60+
{
61+
return true;
62+
}
63+
64+
@Override
65+
public boolean doRunLocal()
66+
{
67+
return false;
68+
}
69+
70+
@Override
71+
public SequenceOutputProcessor getProcessor()
72+
{
73+
return new CellBenderCiteSeqHandler.Processor();
74+
}
75+
76+
public class Processor implements SequenceOutputProcessor
77+
{
78+
@Override
79+
public void init(JobContext ctx, List<SequenceOutputFile> inputFiles, List<RecordedAction> actions, List<SequenceOutputFile> outputsToCreate) throws UnsupportedOperationException, PipelineJobException
80+
{
81+
if (inputFiles.size() > 1)
82+
{
83+
throw new PipelineJobException("Expected a single input");
84+
}
85+
86+
File h5 = getH5(inputFiles.get(0).getFile());
87+
if (!h5.exists())
88+
{
89+
throw new PipelineJobException("Unable to find file: " + h5.getPath());
90+
}
91+
}
92+
93+
private File getH5(File matrix)
94+
{
95+
return new File(matrix.getParentFile().getParentFile(), "raw_feature_bc_matrix.h5");
96+
}
97+
98+
@Override
99+
public void processFilesOnWebserver(PipelineJob job, SequenceAnalysisJobSupport support, List<SequenceOutputFile> inputFiles, JSONObject params, File outputDir, List<RecordedAction> actions, List<SequenceOutputFile> outputsToCreate) throws UnsupportedOperationException, PipelineJobException
100+
{
101+
102+
}
103+
104+
@Override
105+
public void processFilesRemote(List<SequenceOutputFile> inputFiles, JobContext ctx) throws UnsupportedOperationException, PipelineJobException
106+
{
107+
File inputH5 = getH5(inputFiles.get(0).getFile());
108+
File outputH5 = new File(ctx.getOutputDir(), FileUtil.getBaseName(inputH5.getName()) + ".cellbender.h5");
109+
110+
String exe = SequencePipelineService.get().getExeForPackage("CELLBENDERPATH", "cellbender").getPath();
111+
List<String> args = new ArrayList<>(Arrays.asList(
112+
exe, "--input", inputH5.getPath(),
113+
"--output", outputH5.getPath()
114+
));
115+
args.addAll(getClientCommandArgs(ctx.getParams()));
116+
117+
new SimpleScriptWrapper(ctx.getLogger()).execute(args);
118+
if (!outputH5.exists())
119+
{
120+
throw new PipelineJobException("Missing file: " + outputH5.getPath());
121+
}
122+
123+
File filteredH5 = new File(outputH5.getPath().replaceAll(".h5$", "_filtered.h5"));
124+
if (!filteredH5.exists())
125+
{
126+
throw new PipelineJobException("Missing file: " + filteredH5.getPath());
127+
}
128+
129+
SequenceOutputFile so = new SequenceOutputFile();
130+
so.setReadset(inputFiles.get(0).getReadset());
131+
so.setLibrary_id(inputFiles.get(0).getLibrary_id());
132+
so.setFile(filteredH5);
133+
if (so.getReadset() != null)
134+
{
135+
so.setName(ctx.getSequenceSupport().getCachedReadset(so.getReadset()).getName() + ": CellBender Filtered");
136+
}
137+
else
138+
{
139+
so.setName(inputFiles.get(0).getName() + ": CellBender Filtered");
140+
}
141+
so.setCategory("CellBender Filtered CITE-Seq Counts");
142+
ctx.addSequenceOutput(so);
143+
144+
File aggregates = new File(inputH5.getParentFile(), "antibody_analysis/aggregate_barcodes.csv");
145+
if (!aggregates.exists())
146+
{
147+
throw new PipelineJobException("Missing file: " + outputH5.getPath());
148+
}
149+
150+
try
151+
{
152+
File aggregatesCopy = new File(ctx.getOutputDir(), aggregates.getName());
153+
if (aggregatesCopy.exists())
154+
{
155+
aggregatesCopy.delete();
156+
}
157+
158+
FileUtils.copyFile(aggregates, aggregatesCopy);
159+
}
160+
catch (IOException e)
161+
{
162+
throw new PipelineJobException(e);
163+
}
164+
}
165+
}
166+
}

‎singlecell/src/org/labkey/singlecell/run/CellBenderLoupeHandler.java‎

Lines changed: 135 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,135 @@
1+
package org.labkey.singlecell.run;
2+
3+
import org.json.JSONObject;
4+
import org.labkey.api.module.ModuleLoader;
5+
import org.labkey.api.pipeline.PipelineJob;
6+
import org.labkey.api.pipeline.PipelineJobException;
7+
import org.labkey.api.pipeline.RecordedAction;
8+
import org.labkey.api.sequenceanalysis.SequenceOutputFile;
9+
import org.labkey.api.sequenceanalysis.pipeline.AbstractParameterizedOutputHandler;
10+
import org.labkey.api.sequenceanalysis.pipeline.SequenceAnalysisJobSupport;
11+
import org.labkey.api.sequenceanalysis.pipeline.SequenceOutputHandler;
12+
import org.labkey.api.sequenceanalysis.pipeline.SequencePipelineService;
13+
import org.labkey.api.sequenceanalysis.run.SimpleScriptWrapper;
14+
import org.labkey.api.util.FileUtil;
15+
import org.labkey.singlecell.SingleCellModule;
16+
17+
import java.io.File;
18+
import java.util.ArrayList;
19+
import java.util.Arrays;
20+
import java.util.List;
21+
22+
public class CellBenderLoupeHandler extends AbstractParameterizedOutputHandler<SequenceOutputHandler.SequenceOutputProcessor>
23+
{
24+
public CellBenderLoupeHandler()
25+
{
26+
super(ModuleLoader.getInstance().getModule(SingleCellModule.class), "Run CellBender (RNA-seq)", "This will run cellbender on the input cellranger folder and create a subset matrix with background/ambient noise removed.", null, CellBenderCiteSeqHandler.getParams());
27+
}
28+
29+
@Override
30+
public boolean doSplitJobs()
31+
{
32+
return true;
33+
}
34+
35+
@Override
36+
public boolean canProcess(SequenceOutputFile o)
37+
{
38+
return CellRangerGexCountStep.LOUPE_CATEGORY.equals(o.getCategory()) & o.getFile().getName().endsWith("cloupe.cloupe");
39+
}
40+
41+
@Override
42+
public boolean doRunRemote()
43+
{
44+
return true;
45+
}
46+
47+
@Override
48+
public boolean doRunLocal()
49+
{
50+
return false;
51+
}
52+
53+
@Override
54+
public SequenceOutputProcessor getProcessor()
55+
{
56+
return new CellBenderLoupeHandler.Processor();
57+
}
58+
59+
public class Processor implements SequenceOutputProcessor
60+
{
61+
@Override
62+
public void init(JobContext ctx, List<SequenceOutputFile> inputFiles, List<RecordedAction> actions, List<SequenceOutputFile> outputsToCreate) throws UnsupportedOperationException, PipelineJobException
63+
{
64+
if (inputFiles.size() > 1)
65+
{
66+
throw new PipelineJobException("Expected a single input");
67+
}
68+
69+
File h5 = getH5(inputFiles.get(0).getFile());
70+
if (!h5.exists())
71+
{
72+
throw new PipelineJobException("Unable to find file: " + h5.getPath());
73+
}
74+
}
75+
76+
private File getH5(File loupe)
77+
{
78+
return new File(loupe.getParentFile(), "raw_feature_bc_matrix.h5");
79+
}
80+
81+
@Override
82+
public void processFilesOnWebserver(PipelineJob job, SequenceAnalysisJobSupport support, List<SequenceOutputFile> inputFiles, JSONObject params, File outputDir, List<RecordedAction> actions, List<SequenceOutputFile> outputsToCreate) throws UnsupportedOperationException, PipelineJobException
83+
{
84+
85+
}
86+
87+
@Override
88+
public void processFilesRemote(List<SequenceOutputFile> inputFiles, JobContext ctx) throws UnsupportedOperationException, PipelineJobException
89+
{
90+
File filteredH5 = runCellBender(inputFiles.get(0).getFile(), ctx);
91+
92+
SequenceOutputFile so = new SequenceOutputFile();
93+
so.setReadset(inputFiles.get(0).getReadset());
94+
so.setLibrary_id(inputFiles.get(0).getLibrary_id());
95+
so.setFile(filteredH5);
96+
if (so.getReadset() != null)
97+
{
98+
so.setName(ctx.getSequenceSupport().getCachedReadset(so.getReadset()).getName() + ": CellBender Filtered");
99+
}
100+
else
101+
{
102+
so.setName(inputFiles.get(0).getName() + ": CellBender Filtered");
103+
}
104+
so.setCategory("CellBender Filtered RNA-Seq Counts");
105+
ctx.addSequenceOutput(so);
106+
}
107+
108+
public File runCellBender(File input, JobContext ctx) throws PipelineJobException
109+
{
110+
File inputH5 = getH5(input);
111+
File outputH5 = new File(ctx.getOutputDir(), FileUtil.getBaseName(inputH5.getName()) + ".cellbender.h5");
112+
113+
String exe = SequencePipelineService.get().getExeForPackage("CELLBENDERPATH", "cellbender").getPath();
114+
List<String> args = new ArrayList<>(Arrays.asList(
115+
exe, "--input", inputH5.getPath(),
116+
"--output", outputH5.getPath()
117+
));
118+
args.addAll(getClientCommandArgs(ctx.getParams()));
119+
120+
new SimpleScriptWrapper(ctx.getLogger()).execute(args);
121+
if (!outputH5.exists())
122+
{
123+
throw new PipelineJobException("Missing file: " + outputH5.getPath());
124+
}
125+
126+
File filteredH5 = new File(outputH5.getPath().replaceAll(".h5$", "_filtered.h5"));
127+
if (!filteredH5.exists())
128+
{
129+
throw new PipelineJobException("Missing file: " + filteredH5.getPath());
130+
}
131+
132+
return filteredH5;
133+
}
134+
}
135+
}

0 commit comments

Comments
 (0)