Martin DAVY1, Patrick AMAR1,2, Franck Molina1

1 Synthetic & Systems Biology group, Sys2diag UMR 9005 CNRS / ALCEN
2 Bioinfo group, LISN, Univ. Paris Saclay, UMR 9015 CNRS

Download skillFinder
    Windows Version (February 22, 2024)
    Linux (x86_64) Version (February 22, 2024)
    Mac OS X (Catalina) Version (February 22, 2024)
    Source code (February 22, 2024)

Download DATA
    Set of enzymatic reactions from BRENDA (november 08, 2023)
    Set of enzymatic inhibitions from BRENDA (november 08, 2023)

Overview

SkillFinder is a software tool that extracts from a set of enzymatic reactions, those that connect given substrates to a product. Kinetics constants and pH, can be specified to refine the search. The ability of metabolites to inhibit reactions can be selected.

Finding all the paths, from input to output nodes in a directed graph, is known to be an exponential problem. To reduce the computing time, the search is carried out starting both from the inputs to intermediate metabolites, and backward from the output to intermediate metabolites. This way, the depths of searches are halved and the number of nodes to explore is exponentially reduced.

We used BRENDA (https://www.brenda-enzymes.org) for our benchmarks mainly because of the large number of reactions described, and because kinetics constants, pH and inhibitors are also included.

It is used along with Silicell Maker to build the initial biochemical network used to find enzymatic logic gates.

Quick skillFinder user manual

Command line options

Usage: skillFinder -in [inputs] -out [outputs] -br [reactions file] -bi [inhibitions file] -o [output file] [OPTIONS]
 
Mandatory parameters
 
-in "i1;i2;..."input metabolites, separated by ';'
-br [file]input metabolic network file
-bi [file]input file containing the enzymatic inhibitions
-o [file]output file (in SBML format) containing the reactions extracted
 
Options
 
-out "o1;o2;..."output metabolites, separated by ';'
-d numexact distance between input and output metabolites (1 to 4, default to 2).
-sdSearch for reactions at distances smaller or equal to d
-km num:numfilter on Km values by defining an interval (values in mM)
-kcat num:numfilter on kcat values by defining an interval
-ph num:numfilter on pH values by defining an interval
-ki num:numfilter on Ki values by defining an interval (values in mM)
-ic50 num:numfilter on IC50 values by defining an interval (values in mM)
-l [file]generate a file in TSV format containing the list of reactions found
-ecUse EC number instead of enzyme name in XML output file
-h, --helpprint this help
-v, --verboseprint informations during the execution of SkillFinder
 
Examples

Example 1

Inputs: acetone and D-glucose
Output: resorufin

Find all the reactions that connect glucose and acetone inputs to resorufin. The distance between inputs and output is set to 4. The pH of the reactions is between 6 and 8, the Ki of the inhibitions is less than 0.001.

skillFinder -in "acetone;D-glucose" \
            -out "resorufin" \
            -br brendaReaction.gz \
            -bi brendaInhibition.gz \
            -ph 6:8 -ki 0:0.001 \
            -o gluc_acetone-resorufin.xml \
            -v -d 4
9350 reactions are extracted from the 300,000 in the BRENDA database, and written to the output SBML file in a few seconds.

Example 2

Inputs: lactose and ABTSreduced
Output: ABTSoxidized

Find all the reactions that connect ABTSreduced and lactose inputs to ABTSoxidized. The distance between inputs and output is set to 4. The pH of the reactions is between 6 and 8, the Ki of the inhibitions is less than 0.001.

In the command line below, the full name of ABTS is used.

skillFinder -in "reduced 2,2'-azino-bis-(3-ethylbenzthiazole-6-sulfonic acid);lactose" \
            -out "oxidized 2,2'-azino-bis-(3-ethylbenzthiazole-6-sulfonic acid)" \
            -br brendaReaction.gz -bi brendaInhibition.gz \
            -ph 6:8 -ki 0:0.001 \
            -o ABTSred_lactose-ABSTox.xml \
            -v -d 3

23 reactions extracted from the 300,000 in the BRENDA database and written to the output SBML file in a few seconds.

The NetGate plugin of Silicell Maker, using these 23 reactions as input network, has found 3328 enzymatic logic gates. One of them is ABTSoxidized = lactose ∧ ABTSreduced which was tested experimentally in Rialle et al.