Partial correlation simply means computing the correlation between two variables while controlling for all other variables. Partial correlation can be better at detecting associations that represent true dependency, rather than spurious connections that arise from the multicollinearity typically present in 'omics datasets.
In our experience, partial correlation tends to create a sparse network, with much fewer connections between nodes. This can be helpful in reducing the “hairball effect”. More details …
See the image below for an example generated by OmicsAnalyst
