How were the gene node2vec representations trained?

[WuSht]

Hello,
Brilliant JoB! Only, I have a little question:

In Supplementary Note 2: Feature engineering:

the Gene features are coded as Vector representations for L1000 genes are generated from the human protein-protein interaction network extracted from STRING database using node2vec method. The length of these vectors is set at 128, but what is the training objective of the node2vec method, based on some of their properties? Could you put them in detail?

Very very Thank you!

Looking for your reply!