KEGG probiotics

The goal of kegg probiotics is to leverage the data available in the KEGG database to make finding bacterial genomes with potential to degrade carbohydrates of interest while metabolising neuroactive compounds an easier task. To do so, here we share the script find_probiotic_candidates.R, which was successfully used to produce the in-silico analysis conducted in the published article: Paper name.

Overview

The scripts starts from a list of KEGG compound IDs of carbohydrates (or internal carbohydrate subunits of a polymer), and performs the following steps using the information stored in KEGG:

1. Connects carbohydrate’s compounds IDs to IDs of reactions where it participats
2. Connects the reaction IDs to Enzymes (EC codes). If EC codes are provided as an additional optional input for some carbohydrates, those EC numbers are integrated into the next step.
3. Connects EC numbers to Kegg Orthologs (KOs)
4. Connects KOs to all bacterial genomes where that KO was found.

Input example

The scripts takes a tabular input data frame in the following form:

carbohydrate_name	carbohydrate_kegg_id	enzyme
1-Kestose	C03661	NA
Arabinoxylan‐oligosaccharides (AXOS) : Arabinoxylan	C01889	3.2.1.55
Beta-glucan	C00551	NA
Chitooligosaccharides (COS) : Chitositan	C00734	NA
Fructooligosaccharide : Sucrose	C00089	NA
Fructooligosaccharide : Inuline	C08659	3.2.1.7

where:

• carbohydrate_name Identifies the target carbohydrate in an human-friendly way (not used in the script) [optional colum].
• carbohydrate_kegg_id ID of the carbohydrate (or a subunit within a polymer) used in the KEGG db for its identification.
• enzyme ECs of enzymes known to interact with the carbohydrate of interest [optional column].

Output example

There are two main outputs produced by the script.

1. A tabular output with the bacterial genomes showing enzymes capable of degrading the carbohydrates of interest

carbohydrate_name	carbohydrate_kegg_id	rx	definition	ec	ko	abbreviation	t_code	organism	taxonomy
Arabinoxylan‐oligosaccharides (AXOS) : Arabinoxylan	C01889	NA	NA	3.2.1.55	K01209	esa	T00586	Cronobacter sakazakii ATCC BAA-894	prokaryotes_bacteria_gammaproteobacteria_enterobacteria_cronobacter_sakazakii_atcc_baa-894
Arabinoxylan‐oligosaccharides (AXOS) : Arabinoxylan	C01889	NA	NA	3.2.1.55	K01209	csk	T02119	Cronobacter sakazakii ES15	prokaryotes_bacteria_gammaproteobacteria_enterobacteria_cronobacter_sakazakii_es15
Chitooligosaccharides (COS) : Chitositan	C00734	R02333	Chitin + n H2O <=> Chitosan + n Acetate	3.5.1.41	K01452	bcr	T00799	Bacillus cereus AH187	prokaryotes_bacteria_firmicutes_bacilli_bacillus_cereus_ah187
Chitooligosaccharides (COS) : Chitositan	C00734	R02333	Chitin + n H2O <=> Chitosan + n Acetate	3.5.1.41	K01452	btk	T00179	Bacillus thuringiensis serovar konkukian 97-27	prokaryotes_bacteria_firmicutes_bacilli_bacillus_thuringiensis_serovar_konkukian_97-27
Fructooligosaccharide : Sucrose	C00089	R00801	Sucrose + H2O <=> D-Fructose + D-Glucose	3.2.1.10	K01182	kpe	T00772	Klebsiella variicola 342	prokaryotes_bacteria_gammaproteobacteria_enterobacteria_klebsiella_variicola_342
Fructooligosaccharide : Sucrose	C00089	R00801	Sucrose + H2O <=> D-Fructose + D-Glucose	3.2.1.10	K01182	kpk	T03368	Klebsiella variicola KP5-1	prokaryotes_bacteria_gammaproteobacteria_enterobacteria_klebsiella_variicola_kp5-1

2. A tabular output with the bacterial genomes showing enzymes involved in the metabolism of neuroactive compounds

gbm	prokaryotes_bacteria_bacteria_incertae_sedis_bacteria_candidate_phyla_division_tm6_bacterium_gw2011_gwf2_28_16	prokaryotes_bacteria_bacteria_incertae_sedis_chazhemtobacterium_chazhemtobacterium_aquaticus	prokaryotes_bacteria_firmicutes_clostridia_alkaliphilus_oremlandii	prokaryotes_bacteria_firmicutes_others_veillonella_rodentium	prokaryotes_bacteria_firmicutes_others_sporanaerobacter_sp_jn_28	prokaryotes_bacteria_firmicutes_clostridia_aminipila_terrae
MGB004|Kynurenine synthesis	0	0	0	0	0	0
MGB005|Tryptophan synthesis	0	0	0	1	0	0
MGB006|Glutamate synthesis I	0	0	0	1	1	1
MGB007|Glutamate synthesis II	0	0	1	1	0	1
MGB008|Acetylcholine synthesis	0	0	0	0	0	0
MGB009|Histamine synthesis	0	0	0	0	0	0

Comments on the usage of the software

While the script doesn’t need the user to provide a list of already known enzymes that can interact with carbohydrates of interest, adding them can increase the performance of the script and make the search of suitable candidates more thorough. Although the curation of such a list is project specific, there are some resources the authors of this repository find useful:

• Supplementary table 1 from the article Metatranscriptomic holobiont analysis of carbohydrate-active enzymes in the millipede Telodeinopus aoutii (Diplopoda, Spirostreptida)
• Supplementary table 1 & 2 from the article The microbiome’s fiber degradation profile and its relationship with the host diet
• Fiber_Enzyme_matrix.csv from the Github repository associated with the above-mentioned article.

Finally, the script here provided find_probiotic_candidates.R uses the R package omixerRpm which is an R interface to the software Omixer-RPM, which is distributed under an academic Non-commercial Software License Agreement for Omixer-RPM. Details on the conditions of that licence can be found in the software’s GitHub account.