GSEA job errors

[Candice]

Hello,

I am trying to run the following jobs: 1562233 and 1562237. Both are returning different stderr messages. For reference, I have run jobs in GSEA successfully before, so I don't think this is a file format issue since I have checked the formats extensively.

One stderr message is: 

xtools.api.param.BadParamException: Too few samples in the dataset to use this metric
	at edu.mit.broad.genome.alg.DatasetStatsCore.calc2ClassCategoricalMetricMarkerScores(DatasetStatsCore.java:238)
	at edu.mit.broad.genome.alg.markers.PermutationTestBuilder.<init>(PermutationTestBuilder.java:93)
	at edu.mit.broad.genome.alg.gsea.KSTests.shuffleTemplate_canned_templates(KSTests.java:358)
	at edu.mit.broad.genome.alg.gsea.KSTests.shuffleTemplate(KSTests.java:289)
	at edu.mit.broad.genome.alg.gsea.KSTests.executeGsea(KSTests.java:152)
	at edu.mit.broad.genome.alg.gsea.KSTests.executeGsea(KSTests.java:126)
	at xtools.gsea.AbstractGsea2Tool.execute_one(AbstractGsea2Tool.java:101)
	at xtools.gsea.AbstractGsea2Tool.execute_one_with_reporting(AbstractGsea2Tool.java:135)
	at xtools.gsea.Gsea.execute(Gsea.java:114)
	at xtools.api.AbstractTool.tool_main(AbstractTool.java:620)
	at xtools.gsea.Gsea.main(Gsea.java:136)
	at org.genepattern.modules.gsea.GseaWrapper.main(GseaWrapper.java:418)The 

I believe this is saying that 5 samples is insufficient to perform the analysis.

The other stderr message is: 

java.lang.NullPointerException
	at edu.mit.broad.genome.alg.DatasetStatsCore.calc2ClassCategoricalMetricMarkerScores(DatasetStatsCore.java:241)
	at edu.mit.broad.genome.alg.markers.PermutationTestBuilder.<init>(PermutationTestBuilder.java:93)
	at edu.mit.broad.genome.alg.gsea.KSTests.shuffleTemplate_canned_templates(KSTests.java:358)
	at edu.mit.broad.genome.alg.gsea.KSTests.shuffleTemplate(KSTests.java:289)
	at edu.mit.broad.genome.alg.gsea.KSTests.executeGsea(KSTests.java:152)
	at edu.mit.broad.genome.alg.gsea.KSTests.executeGsea(KSTests.java:126)
	at xtools.gsea.AbstractGsea2Tool.execute_one(AbstractGsea2Tool.java:101)
	at xtools.gsea.AbstractGsea2Tool.execute_one_with_reporting(AbstractGsea2Tool.java:135)
	at xtools.gsea.Gsea.execute(Gsea.java:114)
	at xtools.api.AbstractTool.tool_main(AbstractTool.java:620)
	at xtools.gsea.Gsea.main(Gsea.java:136)
	at org.genepattern.modules.gsea.GseaWrapper.main(GseaWrapper.java:418)

Your help is much appreciated!

 

[David Eby]

HI,

Yes, that error message indicates that there are too few samples to use the chosen metric; in particular, the default Signal2Noise metric requires at least three samples per phenotype.

See this section of our User Guide for more details.   You should consult your local statistician to choose the best metric for your data.

If you're using the default parameters, there's a follow-up issue that's likely to come up just beyond this.  You'll need to change the Permutation Type parameter to 'gene_set' as phenotype permutation (the preferred method) is not available with fewer than 7 samples per phenotype.

More details are on this page in our User's Guide.

Regards,

David

David Eby
Consultant
Cancer Informatics Development
Broad Institute of MIT and Harvard
415 Main St, Cambridge, MA 02142, USA
http://www.broadinstitute.org/cancer
http://www.gsea-msigdb.org
https://twitter.com/GSEA_MSigDB
https://twitter.com/GenePattern

--
You received this message because you are subscribed to the Google Groups "GenePattern Help Forum" group.
To unsubscribe from this group and stop receiving emails from it, send an email to genepattern-help+unsubscribe@googlegroups.com.
To post to this group, send email to genepattern-help@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/genepattern-help/8f4df219-b9cd-4c86-ab42-eabcb70801cc%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Candice]

Thank you, this fixed my problem!

To unsubscribe from this group and stop receiving emails from it, send an email to genepattern-he...@googlegroups.com.
To post to this group, send email to genepatt...@googlegroups.com.