Genomic read sequences from clinical exome samples of four (4) pediatric patients with IBD phenotype were processed. IBD patients with acronyms IBD1, IBD3 exhibit Crohn’s disease phenotype, while IBD2 and IBD4 patients display respectively ulcerative colitis and recto-colitis phenotypes ( Table 1 ). Age of analyzed onset inflammatory bowel disease patients ranged from four (4) to 15 years. Indeed, 50% of IBD patients in this study were female and 50% male. In addition, IBD patients’ blood samples, used in the present study were collected from February to August 2018 in reference center and shipped to our laboratory for clinical exome sequencing analysis [6] . DNA library preparation and clinical exome sequencing have been performed following the Illumina MiSeq sequencer manufacture.

We executed a quality control of genomic read sequences obtained from DNA clinical exome sequencing by running Fast-Q quality control package in R programming environment. Next, we performed genomic reads sequences alignment on hg19 human genome running Bowtie 2 package in Galaxy bioinformatics platform by setting standard parameters. We selected genomic read sequence with length ranking between 290 - 300 bp as well as exhibiting quality control score threshold ≥30 for the subjacent bioinformatics as well as genomic and computational statistical analysis ( Figure S1 and Figure S2 ). We retrieved and characterized genetic variants in term of single nucleotide polymorphism (SNP) for each IBD patient by processing BAM (compressed binary files storing sequence reads) files and their indexes from the process of genomic read sequences alignment to hg19 human genome by running in frame the following bioinformatics packages and script (see below) in the Galaxy platform as following:

We carried out genetic variants calling analysis (VCF = Variants Calling Format) from the BAM files and their indices resulting from aligned genomic read sequences to hg19 reference genome. The bioinformatics packages used for the genetic variant calling analysis (VCF analysis) are as following Freebyes [19] , VCF allelic-primitive, SNPEff-Eff [20] , VCF ToTab-delimited, SNPSift-Extract-field, Gemini load [21] , Gemini database-info and Gemini query. Of note, read sequences quality control as well as alignment process, genetic variant calling analysis and their characterization were performed on Galaxy platform. Output files of this analysis are variant calling format files (VCF). Of note, Figure 1 describes and summarizes experimental protocol of bioinformatics and genomic analysis.

As a prelude to the statistical analyses, functional genomic data resulting regarding SNP retrieved from analyzed IBD patients were characterized and structured. Here, we were interested in genetic variants covered by at least 20 genomic read sequences. Indeed, SNP were classified in term of intronic and exonic mutations. Genomic functions regarding IBD patients SNP features are as following:

1) intronic mutation function: 3’ and 5’ un-transcribed regions (3’ and 5’ UTR), downstream gene variant, inter-genic region, intronic variant, upstream gene variant and splice region;

2) Exonic mutation function: induction of the initiator codon, nonsense mutation, loss of the stop codon, loss of the initiator codon, splice mutation and silent mutation.

We checked and quantified genomic functions regarding significant SNP retrieved from inflammatory bowel disease patients by processing data from previous obtained VCF file. Statistical analyses were performed in R programming environment (version 3.6.2). Statistical descriptive analyses included Venn-Diagram, limma, and density plot scripts with the purpose to characterize and assess genetic variants distribution in IBD patient population. We developed a script for endogenous variable assignment of variance for both selected IBD risk factor and pathogenic genetic variants. Analytical statistical analysis included the following statistical tests: i) Kruskal Wallis non-parametric test for variance analysis [22] ; ii) Shapiro normality test; iii) Multiple pairwise comparison analysis (Turkey’s statistical test); iv) Fisher exact test as well as v) Bartlett non-parametric test. Indeed, from the output of the Kruskal-Wallis test, we checked for a significant difference between groups. A significant Kruskal-Wallis test is generally followed up by Dunn’s test to identify which groups are different. A test is considered significant at α threshold = 0.05 (p ≤ 0.05). Genomic functions were assigned to retrieve pathogenic and risk factor SNP genetic variants by using ENSEMBL genomic database. The statistical parameter Fitcon, which indicates the probability of a given mutation having a significant impact on the phenotype, has been introduced to characterize IBD patients CD and UC phenotypes. The following Fitcon parameter values have been set for the: i) intronic genetic variants: risk factors and/or pathogens: 0.3 ≤ Fitcon ≤ 0.4 and ii) exonic genetic variants: risk and/or pathogenic factors: Fitcon ≥ 0.6.

We checked for IBD patients’ genomic read quality control. Genomic read sequences quality Phred score is higher than 30 ( Figure S1 ). Exome DNA genomic read sequences size raking between 100 - 300 bp ( Figure S2 ). According to Figure S2 , the majority of genomic sequences for the four analyzed IBD patients have a size of 300 bp. The MiSeq Illumina sequencer carried out sequencing experiment. The total number of obtained genomic read sequences of considered IBD patients is ranking between 1,830,550 - 2,128,442 sequences. Of note, the average rate of those sequences was estimated to 1,958,395 with a variation coefficient (CV) = 7.06%. Interestingly, more than 70% of aligned genomic read sequences result to be specific for their gene target, while 9% of them claim to recognize more than one gene model targets ( Table 2 ). Taking together, 80% of IBD patients genomic read sequences have been aligned on the h19 human genome ( Table 2 ). Of note, approximately 20% of the genomic read sequences have been not aligned on human h19 reference genome ( Table 2 ). We performed a multivariate comparative statistical analysis regarding IBD patients basing on the proportion of aligned and/or not aligned read sequences, aiming to assess read sequences distribution on h19 human genome (reference genome) in prelude to genetic variant calling as well as to subjacent functional genomic analysis. Multivariate comparative analysis shown no variance difference (p = 0.87) between the four analyzed IBD patients ( Figure 2(A) ). In the other words, inflammatory disease patients 1, 2, 3 and 4 exhibited high variance homogeneity considering aligned genomic read sequences of considered IBD pediatric patients on the h19 human genome ( Figure 2(B) ). Considering as a whole, IBD patients exhibited homogeneous features for the purpose of subjacent genetic variance calling and functional genomic analysis.

Analysis revealed 6465 - 8492 genetic variants covered by at least 20 genomic read sequences for all analyzed IBD patients. IBD patient 4 recorded the lowest number of genetic mutations (6465) while IBD patient 3 recorded the highest number (8492) of genetic mutations. IBD patients 1 and 2 respectively reported 7782 and 7924 genetic variants ( Table 3 ). Next, we discriminated the proportions of those genetic variants in term of homozygosity and heterozygosity’s mutation. Findings revealed respectively 4037 and 2444 heterozygous and homozygous mutations in the IBD pediatric patient 1. The same analyzes showed respectively 4212 and 2639 heterozygous and homozygous mutations in the IBD pediatric 2. IBD pediatric patient 3 recorded respectively 4650 and 2722 heterozygous and homozygous mutations. Analysis of the typology of genetic mutation in the IBD pediatric patient 4, revealed respectively 3370 and 2215 heterozygous and homozygous mutation ( Table 3 ). Of note, genetic mutations retrieved in the IBD population, mainly are single nucleotide polymorphisms (SNPs) inducing in some cases amino acid changing. Thus, we recorded 3669, 3829, 4054 and 3347 amino acid changes respectively in IBD pediatric patients 1, 2, 3 and 4 ( Table 3 ). We stimulated a genomic comparative analysis between IBD patients 1, 2, 3 and 4 basing on heterozygous and homozygous genetic variants inducing amino acid changes. Turkey’s statistical test revealed a non-significant difference in term of variability (p > 0.05) between analyzed IBD pediatric patients 1, 2, 3 and 4 ( Figure 3 ). This result would be in favor of the inflammatory bowel disease (IBD) phenotype shared by all of the patients.

We checked for IBD population variability by assessing genomic functions induced by exonic SNP genetic variants. Analysis revealed the following genomic functions: initiator codon induction, nonsense mutation, stop codon loss, initiator codon loss, splice mutation and silent mutation ( Table 4 ). The recurrent genomic functions induces by exonic SNP in the analyzed IBD patients population are nonsense mutations and silent mutations, while the rare genomic functions are introduction of stop codon and gain of initiator codon respectively ( Table 4 ). The rarest genomic functions in the present analysis are represented by i) stop codon loss in IBD patient 4 and ii) introduction of initiator codon in IBD patient 3. Interestingly, genomic analysis revealed a modest frequency of genetic mutations in several splice regions of the genomes. Indeed, Fisher’s exact test showed a significant recurrence of splice mutations in heterozygote in contrast to splice mutations in homozygote (p = 0.04). The same test supported a non-significant difference for the genomic functions nonsense mutations and silent mutations in terms of hetero and homozygosity (p = 0.84). We checked for the normality distribution regarding heterozygous and homozygous mutations inducing amino acid (aa) changes in analyzed IBD population ( Figure 4 ). Shapiro normality test suggested an asymmetric distribution (p = 0.00) of genomic functions associated to heterozygous and homozygous exonic SNP mutation in the IBD population ( Figure 4(A) ). Density plot graphic descriptive analysis, relative to those genomic functions distribution exhibited a similar profile between the four analyzed IBD patients. Kruskal Wallis non-parametric test evaluating variance difference on that IBD population in term of genomic functions distribution (33 genomic functions) revealed no significant variance difference (p = 0.72) between IBD patients 1, 2, 3 and 4 ( Figure 4(B) ). Interestingly, estimated eta-squared of the Kruskal Wallis test effect size suggested small effect (0.01 - <0.06) of above retrieved genomic functions distribution in affecting IBD population variability.

IC = Initiator codon; MM = Missense mutation; SpM = Splice mutation; ICL = Initiator codon loss; ICG = Initiator codon gain; SCI = Stop codon introduction; SCL = Stop codon loss; SM = Silent mutation, IC = Initiator codon; MM = Missense mutation; SpM = Splice mutation; ICL = Initiator codon loss; ICG = Initiator codon gain; SCI = Stop codon introduction; SCL = Stop codon loss; SM = Silent mutation.

We analyzed the impact of intronic SNP genetic mutation on the genetic variability of IBD patients. Before doing that, we checked for exonic SNP distribution normality in the IBD population. Of note, genomic functions in terms of intronic SNPs genetic mutations were observed in the following genome regions: i) 3’ and 5’ non-transcribed regions (3’ and 5’ UTR), ii) downstream gene variant and iii) inter-genic regions, iv) intronic variant, v) upstream gene variant and vi) splice regions ( Table 5 ). The majority of intronic SNPs retrieved in the IBD population are represented by i) intronic mutation variants and ii) mutation in splice and inter-genic regions ( Table 5 ) that exhibited significant difference in term of heterozygous and homozygous mutations (p < 0.05). Thus, intronic gene variants are more frequent in heterozygosity than in homozygosity (p = 0.02). The same analysis suggests a greater number of mutations in heterozygous splice regions compared to those in homozygosity (p = 0.0002). Inter-gene mutations are more frequent in homozygosity (p = 0.002). Mutations in the 3’ and 5’ non-transcribed regions (the 3’UTR and 5’UTR genomic regions) have a more significant average frequency in heterozygosity than in homozygosity (p < 0.05) for each IDB patients ( Table 5 ). Considering, as a whole IBD population seem to exhibit a variability in terms of intronic SNP variants detected in heterozygous and homozygous as opposite to those detected in exonic regions. We checked for the normality distribution of intronic SNP (genomic functions) in the IBD patient’s population. Shapiro normality test revealed asymmetric distribution of intronic SNP (genomic functions) in the IBD population (p < 0.05) ( Figure 5(A) ). Of note, Kruskal-Wallis no-parametric test suggested no variance difference regarding IBD population by processing intronic SNP genomic functions distribution in IBD patients genome (p > 0.05) ( Figure 5(B) ).

RNA message includes untranslated regions upstream (5’UTR) and downstream (3’UTR) of coding sequence; DGV = Downstream Gene Variant; IGR = Inter-genic Region; IV = Intronic Variant; SIC = Sequence variant that changes the non-coding exon sequence in a non-coding transcript; SAS = Splice Acceptor Site (3’ end of intron); SDS = Splice Donor Site (5’ end of intron); SR = Splice Region; UGV = Upstream Gene Variant.

We previously shown a significant variability in IBD pediatric population by comparing heterozygous and homozygous SNPs variants in exonic genomic regions. Analysis revealed variability in terms of heterozygous non-coding region mutation (intronic SNPs) for splice, inter-genic and 3’UTR and 5’UTR genomic regions ( Figure 5(A) ). Herein we characterized IBD population by combining heterozygous and homozygous exonic and intronic SNP variants with the purpose to estimate that population variability. Shapiro normality test suggested random distribution regarding intronic and exonic homozygous and heterozygous SNPs in IBD patients’ genome ( Table 6 ). Of note, all analyzed IBD patients seem to exhibit the same statistic features regarding heterozygous and homozygous intronic and exonic genomic SNPs distribution ( Table 6 , Figure 6(A) ). Interestingly, by removing intronic variants as well as exonic missense and silent SNP mutations, statistical analysis revealed significant variability in the inflammatory bowel disease pediatric patient’s population ( Figure 6(B) and Figure 7 ). Wilcoxon pairwise comparative analysis suggested significant difference between i) IBD patients 1 and 2 (p = 0.04), between ii) IBD patient 1 and 3 (p = 0.02), between iii) IBD patients 2 and 3 (p = 0.03), between iv) IDB patients 3 and 4 (p = 0.02) and between IBD 1 and 4 (p = 0.04) ( Figure 7 and Table S1 ). Of note, intronic and non-coding exonic SNP exhibited a significant normality (p < 0.05) in the IBD patient 1, 2, 3 and 4 as opposite to the other analyzed SNP ( Table S1 ).

We introduced Fitcon parameter with the purpose to reveal homozygote and heterozygote intronic as well as exonic SNPs that significantly influence IBD patient’s genomic functions. Density plot analysis suggested asymmetric distribution of SNP influencing significantly IBD patient genomic functions by introducing Fitcon parameter ( Figure 8(A) ). The same descriptive statistical analysis exhibited similar distribution regarding heterozygote and homozygote SNPs genetic variants selected by Fitcon parameter in the IBD population ( Figure 8(A) ). Of note, hierarchical clustering analysis exhibit an apparent variability in the IBD pediatric population by considering homozygous and heterozygous intronic as well as exonic SNPs mutations that significantly influence IBD patient’s genomic functions ( Figure 8(B) ). Interestingly, estimated eta-squared of the Kruskal Wallis test effect size suggested moderate effect of those SNPs genomic functions distribution in affecting IBD population variability. However, Kruskal Wallis test suggested Fitcon parameter as reducing IBD pediatric population variability by contrast to descriptive result of hierarchical clustering analysis ( Figure S3 ) as well as Kruskal Wallis test effect size ( Figure 8(B) ).

We analyzed the impact of pathogenic mutations and as well risk factor variants of inflammatory bowel diseases on the genetic diversity of a pediatric IBD patient population. Analysis revealed 29 and 64 IBD genetic pathogenic and risk factor variant respectively ( Table 7 and Table S3 and Table S4 ). Of note, Shapiro normality test revealed an abnormal distribution of risk factor (Bandwidth = 0.15) and as well pathogenic (Bandwidth = 0.17) SNPs genetic variants in the IBD population ( Figure S4 ). Bartlett non-parametric test revealed a non-significant variance difference for analyzed IBD patient population by processing IBD genetic risk factor variants (SNPs) (p = 0.99) ( Table S3 ). The same analysis revealed a relative significant variance difference in the IBD patients population by processing IBD pathogenic genetic variants by opposite genetic risk factor (p = 0.28) ( Table S4 ). In addition, finding revealed that more than 95% of IBD pathogenic and risk factor genetic SNP mutations happened in the coding regions. Proportion survey regarding IBD risk factor and genetic pathogenic variants by introducing Fitcon parameter suggested more than 50% of exon mutations regarding IBD genetic pathogenic (55%) and risk factors (69%) displayed a significant probability value (Fitcon ≥ 0.6) in terms of significantly impacting IBD patient population phenotype ( Table 7 ). Interestingly, finding revealed IBD pathogenic and risk factor SNP variants as exhibiting the same performances by introducing the Fitcon parameter in selecting both pathogenic and risk factor variants, confirming Fitcon parameter as normalizing factor in assessing IBD patient phenotype ( Figure 9 ). Of note, genetic variability in IBD pediatric patients population through endogenous variable assignment of variance regarding normalized IBD pathogenic and risk factor variants suggested 2 IBD patients cluster groups as following, group 1 including IBD patients 1 and 3,while group 2 included IBD patients 2 and 4 ( Figure 9 ).

Findings revealed 66 genetic risk factor SNP variants including 14, 4 and 3 genetic risk factor variants associated with metabolic, inflammatory bowel and autoimmune diseases respectively, while 68% of revealed SNPs claim to be associated with other diseases ( Figure 10(A) and Table S3 ). Of note, rs429358 (c.466 T > C/p. (Cys156Arg)) and rs1805097 (c.3170G > A/p. (Gly1057Asp)) genetic variants, respectively from APOE and IRS2 genes associated to metabolic disorder have been detected in IBD patient 1. IBD patient 2 exhibited three specific genetic variants associated to metabolic disorders i.e. rs4880 (c.47T > C/p. (Val16Ala)), rs13266634 (c.973C > T/p. (Arg325Trp)) and rs231775 (c.49A > G/p. (Thr17Ala)) respectively from SOD2, SLC30A8 and CTLA4 genes. rs2904552 (c.1292G > A/p.(Arg431His)) from gene PRODH involved in metabolic abnormalities, rs34911341 (c.152G > A/p.(Arg51Gln)) from GHRL gene, a susceptibility factor for obesity and rs1053874 (c.731 G > A/p.(Arg244Gln)) from DNASE1 gene, a susceptibility factor for systemic lupus erythematosus disease and body fat distribution have been specifically detected in IBD patient 4. The same analysis suggested rs12150220 homozygous variant (c.464 T > A/p. (Leu155His)) from NLRP1 gene involved in Vitiligo disease associated with multiple autoimmune diseases, rs373237 (c.935C > T/p. (Thr312Met)) from CX3CR1 gene and rs3732379 (c.841 G > A/p. (Val281Ile)) variant from CX3CR1 gene were revealed only in IBD patient 4 ( Table S3 ). Of note, analysis showed inflammatory bowel disease genetic risk factors commonly shared in t the IBD pediatric population involved in metabolic, inflammatory bowel and autoimmune disorder. The homozygote genetic variants rs450046 (c.1562 G > A/p.(Arg521Gln)) and rs1799983 (c.894T > G/p.(Asp298Glu)) respectively from genes PRODH and NOS3 and the genetic variant rs237025 (c.163G > A/p.(Val55Met)) from SUMO4 gene as well as risk factors variant, rs180223 (c.2200 T > G/p(.Ser734Ala)) involved in metabolic syndrome, and homozygous genetic variant rs853326 (c.3082A > G/p.(Met1028Val)) from TG gene, associated with thyroid autoimmune diseases and rs2241880 genetic variant (c.898A > G/p.(Thr300Ala)) from ATG16L1 gene a risk factor for Crohn’s disease and characteristic of inflammatory bowel diseases are shared by the four analyzed IBD patients population ( Table S3 ). IBD patients 1, 2 and 3 exhibited homozygous genetic variants rs861539 (c.722C > T/p. (Thr241Met)) and rs1044498 (c.517A > C p. (Lys173Gln)), respectively from XRCC3 and ENPP1 gene, linked to metabolic disorders. IBD patients 2 and 3 share rs1799945 (c.187C > G/p. (His63Asp)) genetic variant of HFE gene, a susceptibility factor for metabolic abnormalities, while homozygous genetic variant rs1131454 (c.484G > A/p. (Gly162Ser)) of OAS1 gene, a susceptibility factor of metabolic abnormalities, is shared by IBD 1 and 3 patients ( Figure 10 and Table S3 ). Analysis revealed homozygous rs5219 (c.67A > G/p. (Lys23Glu)) genetic variant of KCNJ11 gene involved in abnormal metabolic disorders in IBD patients 1, 2 and 3 ( Figure 10 and Table S3 ). IBD patients 2 and 4 share rs2066844 (c.2104C > T/p. (Arg702Trp)) genetic variant of NOD2 gene a Crohn disease risk factor and a well know IBD biomarker ( Figure 10 , Table S3 and Figure S5 ). IBD patients 2, 3 and 4 share homozygous rs7076156 (c.184A > G/p. (Thr62Ala)) genetic variant of ZNF365 gene, a risk factor of Crohn’s disease. IBD patients 1, 3 and 4 share the homozygous rs2227564 (c.422T > C/p. (Leu141Pro)) and rs1169288 (c.79A > C/p. (Ile27Leu)) genetic variants respectively of PLAU and HNF1A genes that are risk factors of Crohn’s disease and metabolic syndrome respectively.

Finding revealed 31 IBD pathogenic genetic variants. Among them, nine (9) pathogenic genetic variants result to be involve in metabolic disorders, while one (1) pathogenic variant is associated to inflammatory bowel disease ( Figure 10(B) ). The same analysis revealed 3 pathogenic genetic variants associated with autoimmunity and 18 pathogenic variant linked to other diseases ( Figure 10(B) ). The analysis revealed heterozygous rs429358 (c.466T > C /p. (Cys156Arg)) pathogenic variant of APOE gene involved in metabolic abnormalities in IBD patient 1. Rs2904552 (c.1292G > A/p. (Arg431His)) pathogenic genetic variants of PRODH gene involved in metabolic abnormalities, rs3732378 (c.935 C > T/p. (Thr312Met)) and rs3732379 (c.841G > A/p. (Val281Ile)) pathogenic genetic variants of CX3CR1 gene, susceptibility factor for acquiring immunodeficiency syndrome, were revealed only in patient inflammatory bowel (IBD) patient 4 ( Table S4 ). IBD patients 1, 2, 3 and 4 share six (6) pathogenic genetic variants ( Figure 10(B) ) and 4 of them i.e. rs450046 homozygous (c.1562G > A/p. (Arg521Gln)), rs820878 homozygous (c.185T > C/p. (Leu62Ser)), rs1169305 homozygous (c.1741 A > G/p. (Ser581Gly)) and rs1799983 homozygous (c.894T > G/p. (Asp298Glu)), are IBD pathogenic variants (SNPs) involved in metabolic pathologies ( Table S4 and Figure 4(B) ). Of note, IBD patients 3 and 4 share the pathogenic variant rs17580 (c.863A > T/p. (Glu288Val)) of SERPINA1 gene associated with the control of low-density lipoprotein cholesterol levels. IBD patients 2 and 3 shared rs1799945 (c.187C > G/p. (His63Asp)) pathogenic variant of HFE gene involved in metabolic disorders. The heterozygous rs10065172 pathogenic genetic variant (c.313C > T/p. (Leu105Leu)) of IRGM gene associated with inflammatory bowel disease was revealed in IBD patients 1, 2 and 4 patients ( Figure S5 ). IBD patients 1, 3 and 4 share the rs10010131 and rs351855 genetic variant of WFS1 and FGFR4 genes respectively susceptible factor of metabolic syndrome and body fat distribution. The pathogenic rs1805010 genetic variant of IL4R gene involved in the acquired immunodeficiency process has been revealed in IBD patients 1, 2 and 4 ( Table S4 and Figure 10(B) ). Of note, statistical analysis evaluating variability in the IBD population by processing IBD pathogenic and as well risk factor genetic variants together highlights high similarity (p > 0.05) between Crohn’s disease (CD) and ulcerative colitis phenotype ( Figure 10(C) ).

Here we embarked in characterizing IBD pediatric patient’s variability by processing separately IBD pathogenic and risk factor SNP genetic variants. Because of IBD pathogenic and risk factor SNP genetic variants displayed asymmetric distribution (Shapiro test, p < 0.05), we performed Kruskal-Wallis test assessing IBD patient population variability ( Figure 11(A) and Figure 11(B) , suggesting no significant variability between the four analyzed IBD pediatric patients ( Figure 11(c) and Figure 11(D) . However, a comparative analysis between IBD pathogenic SNP genetic variants (p = 0.4) and IBD SNP risk factors (p = 0.76) in evaluating IBD population variability, suggested a moderate aptitude of pathogenic genetic variants in categorizing IBD patients phenotype ( Figure 11(C) and Figure 11(D) ). Interestingly Euclidean distance clustering analysis suggested a relative high aptitude of IBD SNP pathogenic genetic variants in categorizing IBD patient’s in ulcerative colitis and Crohn’s disease phenotype as opposite to IBD risk factor variants ( Figure 12(A) and Figure 12(B) ). In the other words, analyzed IBD patients exhibit the same phenotype by considering IBD risk factors as opposite to IBD pathogenic SNP genetic variants, that clustered together i) IBD patients 2 and 4 recognized as exhibiting ulcerative colitis phenotype and ii) IBD patients 1 and 3 patients with Crohn’s disease phenotype ( Figure 12(B) ). This result confirm Fitcon parameter clustering analysis that suggested two IBD patients groups as following; i) IBD patients 1 and 3 and ii) IBD patients 2 and 4 ( Figure 9 ).