Is CORNETO installation incompatible with LIANA?

[jiangzh-coder]

In "Differential Expression Analysis for CCC & Downstream Signalling Networks" part ,
i met following error:
Is CORNETO installation incompatible with LIANA?
########1 i have tried used 2 ways to install corneto:
pip install corneto cvxpy scipy gurobipy or
pip install git+https://github.com/saezlab/corneto.git#egg=corneto[os]
#2 gurobipy i have followed "Differential Expression..." tutorial to install.

Currently my versions are；
pip show liana #Version: 1.4.0
pip show corneto #Version: 1.0.0a0

AttributeError Traceback (most recent call last)
Cell In[21], line 1
----> 1 df_res, problem = li.mt.find_causalnet(
2 prior_graph,
3 input_scores,
4 output_scores,
5 node_weights,
6 # penalize (max_penalty) nodes with counts in less than 0.1 of the cells
7 node_cutoff=0.1,
8 max_penalty=1,
9 # the penaly of those in > 0.1 prop of cells set to:
10 min_penalty=0.01,
11 edge_penalty=0.1,
12 verbose=False,
13 max_runs=50, # NOTE that this repeats the solving either until the max runs are reached
14 stable_runs=10, # or until X number of consequitive stable runs are reached (i.e. no new edges are added)
15 solver='gurobi' # 'scipy' is available by default, but often results in suboptimal solutions
16 )

File ~/miniconda3/envs/omicverse/lib/python3.10/site-packages/liana/method/fun/_causalnet.py:193, in find_causalnet(prior_graph, input_node_scores, output_node_scores, node_weights, node_cutoff, min_penalty, max_penalty, missing_penalty, edge_penalty, solver, seed, max_runs, stable_runs, verbose, **kwargs)
189 c_node_penalties = {k: node_penalties.get(k, missing_penalty) + rng.uniform(min_penalty/20, min_penalty/10)
190 if k not in measured_nodes else 0.0 for k in prior_graph.vertices}
192 _logg("Building CORNETO problem...", verbose=verbose)
--> 193 P, G = cn.methods.carnival._extended_carnival_problem(
194 prior_graph,
195 input_node_scores,
196 output_node_scores,
197 node_penalties=c_node_penalties,
198 edge_penalty=edge_penalty
199 )
201 # E is the variable with 1 if edge activates or inhibits, 0 otherwise
202 E = P.symbols['reaction_sends_activation_c0'] + P.symbols['reaction_sends_inhibition_c0']

AttributeError: module 'corneto.methods.carnival' has no attribute '_extended_carnival_problem'

[dbdimitrov]

pip install corneto==0.9.1-alpha.5 this should make it work :)

[jiangzh-coder]

Thank you vey much!

[Cladors]

Just saying that I have had the same issue. _extended_carnival_problem does not exists in the corneto.methods.carnival.py file any more. They have done multiple edits to that file in the past year or so. See here: https://github.com/saezlab/corneto/commits/dev/corneto/methods/carnival.py