Question: SystemPipeR GO term enrichment analysis - filterDEG
0
3.5 years ago by
A Gossner50
UK
A Gossner50 wrote:

Unsure what exactly the 'UporDown' value represents in terms of the differrentially expressed genes associated with the corresponding GO term. Does it represent the total number of genes up and down regulated (that pass the filter settings) for that specific GO term and if it so why do you get GO terms that have only 'UporDown' values but no values for only 'down' or 'up'?

systempiper go term • 560 views
modified 3.5 years ago • written 3.5 years ago by A Gossner50
Answer: SystemPipeR GO term enrichment analysis - filterDEG
0
3.5 years ago by
Thomas Girke1.7k
United States
Thomas Girke1.7k wrote:

This could happen when a gene set is smaller than the minimum number defined under the CLSZ argument of the GOCluster_Report function, or the enrichment p-value for a GO node does not meet the threshold requirement defined under cutoff. Note, this function knows nothing about the nature or source of the gene sets you are passing on. It simply allows you to process many gene sets stored in a list of character vectors where the set names (here names of list components) are maintained in the report.

Thomas

Okay, Thank you.

 CLID CLSZ GOID NodeSize SampleMatch Phyper Padj Term Ont GO_OL_Match 2815 M-C_up_down 179 GO:0006954 294 13 1.20E-05 0.001288505 inflammatory response BP 53 24071 M-C_up 121 GO:0006457 139 8 5.20E-06 0.000395476 protein folding BP 8 187211 M-C_up 121 GO:0070488 2 2 4.72E-05 0.003584582 neutrophil aggregation BP 36 194611 M-C_up 121 GO:0071621 60 5 5.73E-05 0.00435455 granulocyte chemotaxis BP 2 132691 M-C_up 121 GO:0045454 62 5 6.72E-05 0.005103483 cell redox homeostasis BP 5 15403 M-C_up 121 GO:0050873 37 4 0.000118504 0.009006332 brown fat cell differentiation BP 4

These BP results meet CLSZ and cutoff (Padj) but GO:0006954 only up_down?

You are not saying what your cutoff for the adjusted p-value was. Note, for GO:0006954 the adjusted p-value is barely significant. Minor changes in the enrichment may bring it above the cutoff you are using, so I wouldn't be surprised if any subset of up_down was dropped in the result. If you want to force the function to return all result then set the cutoff to 1. By default this is avoided since when testing many sample sets this will return a very large table with hundred thousands of rows.

Thomas