Hello everyone, i am working on network analysis of Differentially Expressed Genes, but the problem i am facing is that, the gene set is very short (of around 16 genes only), so when i'm trying to build an interaction using these genes in STRING webtool, the nodes starts to dissociate from the interaction when i set the confidence score to 0.400 or above (as the genes are related to Homo sapiens). So, to build an interaction between these genes, i've to decrease the confidence score to 0.150. At this score only i'am getting an interaction. So, i want to ask whether this low confidence score is valid for this short gene set related to Homo sapiens or else what should i do in this case?
okay, but will it be valid for my study to add more related genes which are not in my desired gene set?
Yes if our objective is to find with which genes they are related to. We use string to map the input geneset on a protein protein interaction network to find neighbours. Hence even if ur genes are 20, u can extend further to find closest and distant neighbours in the PPI network. U can use confidence to infer the interaction strength.
Is 0.150 confidence score (STRING) justified if i am working with a set of less than 20 genes?
There is no hard and fast rule that 0.15 is bad. So u can go with this low threshold and convey your results accordingly. But my suggestion would be to increase the score to lets say 0.4 and add more genes in the network to get a decent size of network. Infer the enrichment analysis with the grown network
Ok, thank you so much, this was very much helpful for me.