Graph tool for assortative communities in signed networks

Question

I have a signed, weighted network where edges are real numbers that can take values in [-1, 1], this is calculated from the Spearman correlations between the nodes. I wish to detect assortative structures only, but where the sum of the positive edges within groups is maximised and the sum of the negative edges between groups is maximised.

In the graph-tool documentation here I can see that there is:

1. The planted partition (PPBlockState) which searches for exclusively assortative communities.
2. The option to include the weights as covariates, either by (a) shifting them to make them all positive, or (b) including absolute value weights and signs as two separate covariates.

If I combine (1) and (2) will the method understand the role of the signs and weights correctly. If so, how does this work? Also, how does shifting (as in 2a) affect the problem?

I am also open to using other (Python or R) packages.