Table 2 Comparison of the pathway interaction method and pathway ORA. Here we summarized a number of differences and similarities between the pathway interaction method and pathway ORA. As can be seen below, they differ in the databases used, number of genes detected, the way significant genes are selected, and the structure of resulting subnetworks. On the other hand, in our study both methods identified the same biological processes. The two have different strengths and weaknesses; we suggest that researchers exploit both methods to gain a bigger picture of the biological mechanism

Database Used

WikiPathways, STRING

WikiPathways

Detected Unit

Significant path (two or more genes connected by protein-protein interaction)

Individual gene

Selection Criteria

Sum of weights

(transformed T-statistics)

P-value and log2FC score

Resulting Subnetwork

Pathway-gene-gene-pathway

Gene-pathway-gene-pathway

Biological Processes Identified for Psychiatric 22q11DS

NK cell function PI3K/Akt signalling

NK cell function PI3K/Akt signalling

Strength

Can detect multiple genes collectively having a strong expression profile, backed by prior knowledge.

Can detect single strongly expressed genes using flexible criteria.

Weakness

Strongly dependent on pre-existing relationships between gene products. Genes with unidentified relationships will be ignored. Detection criteria are hard to adjust or modify.

Individual gene selection can be highly dependent on the significance criteria. Depending on data quality, spurious genes might be mistaken as being significant.