TypeError in expression_profile.py

[jotech]

I encountered an error when calling the imat function:

imat(recon, exprof, (0, 30), condition="cond1")
[...]
TypeError: Call constructor takes at most 3 positional arguments

I can track it down to a sympy call by parse_expr:

./driven/data_sets/expression_profile.py in _map_gene_to_rxn(self, reaction, gene_values, by)
    333         rule = reaction.gene_reaction_rule.replace("and", "*").replace("or",
    334                                                                        "+")
--> 335         expression = parse_expr(rule, local_dict)
    336         if by == "or2max_and2min":
    337             expression = expression.replace(Mul, Min).replace(Add, Max)

A minimal example to reproduce the error would be:

from sympy import Symbol
from sympy.parsing.ast_parser import parse_expr
local_dict = {'5161.0': 5161.0,
 '1738.0': 1738.0,
 '5160.0': 5160.0,
 '1737.0': 1737.0,
 '5162.0': 5162.0,
 '8050.0': 8050.0}
rule = '(1738.0 * 8050.0) * (5161.0 * 5162.0) * (1737.0) + (1738.0 * 8050.0) * (5160.0 * 5162.0) * (1737.0)'
expression = parse_expr(rule, local_dict)
[...]
TypeError: Call constructor takes at most 3 positional arguments

Do you have any clue what's going on here?

extended log file
python 3.7.2+
sympy 1.3

[arabidopsis]

Yes. sympy is using a 5 argument version of ast.Call which was
removed in Py3.5. see https://greentreesnakes.readthedocs.io/en/latest/nodes.html#Call

[arabidopsis]

Here's a monkey patch (might create a pull request if I have time)

from cobra.core.gene import parse_gpr
from sympy import Mul, Add, Symbol, Max, Min
from ast import And, Or, NodeTransformer

class Transform(NodeTransformer):

    def visit_Name(self, node):
        return Symbol(node.id)
    def visit_BoolOp(self, node):
        values = [self.visit(v) for v in node.values]
        if isinstance(node.op, And):
            return Mul(*values)
        if isinstance(node.op, Or):
            return Add(*values)
        return node

def parse_expr(s):
    a, genes = parse_gpr(s.strip())
    a = Transform().visit(a)
    return a.body

def _map_gene_to_rxn(self, reaction, gene_values, by):
    rule = reaction.gene_reaction_rule
    expression = parse_expr(rule)
    if by == "or2max_and2min":
        expression = expression.replace(Mul, Min).replace(Add, Max)
    elif by == "or2sum_and2min":
        expression = expression.replace(Mul, Min)
    return expression.subs(gene_values).evalf()

from driven.data_sets.expression_profile import ExpressionProfile
ExpressionProfile._map_gene_to_rxn = _map_gene_to_rxn

[arabidopsis]

Ok... I created a pull request. Let's see what happens :)

[jotech]

thank you very much it helped indeed :)