Thank you so much! I will try this out.
I am performing GO enrichment analysis on PPI network clusters and encountering high mapping failure in clusterProfiler::bitr(). I would appreciate any clarifications on whether this indicates a flaw in my workflow.
Workflow
Downloaded interaction data from HIPPIE (TAB format).
Filtered interactions with score >= 0.70.
Imported only the protein names (e.g., ITA7, ERBB2) into Cytoscape to build the network.
Selected a predefined gene list in the network and constructed a first-neighbor subnetwork.
Ran MCODE to detect clusters.
Exported the top 5 clusters as CSV files.
Attempted GO enrichment in R using clusterProfiler:
Issue
I receive warnings such as:
Warning: 1: In bitr(gene1, fromType = "SYMBOL", toType = "ENTREZID", OrgDb = org.Hs.eg.db) : 81.13% of input gene IDs are fail to map... 2: In bitr(gene1, fromType = "SYMBOL", toType = "ENTREZID", OrgDb = org.Hs.eg.db) : 41.03% of input gene IDs are fail to map...
The cluster gene names look like this:
RL37A RS16 PRS8.... etc.
Is the mapping failure due to incorrect identifier type selection rather than an error in clusterProfiler?. Should I instead use the Entrez ID columns from the original HIPPIE TAB file for enrichment?
I would appreciate any guidance on the correct way to handle HIPPIE-derived identifiers for downstream GO/KEGG analysis.
1 answer
Hi, Yes it looks like the error is in the identifiers. You have fromType set to "SYMBOL" which expects valid HGNC symbols and RS16 and PRS8 are not valid. When you build the network in Cytoscape either name the nodes by a stable identifer like Entrez IDs or just carry along the Entrez IDs, cluster, and export those Entrez IDs for enrichment analysis.
chris
Log in to answer this question.