*2.1. Study Population*

Patients with ARF undergoing follow-up examination at the Clinic of Thoracic and Cardiovascular Surgery of Fann National University Hospital Centre in Dakar, Senegal, were included herein. The study was approved by the ethics and research committee of Cheikh Anta Diop University (reference number: Protocol 0274/2018/CER/UCAD), and patients provided written informed consent prior to their participation in the study in accordance with the tenets of the Declaration of Helsinki. Some of these patients had undergone valvular replacement surgery, while others did not receive surgical intervention. Healthy individuals were recruited as controls.

Patients were divided into three groups: First group, healthy individuals (control group); second group, unoperated ARF patients; third group, operated ARF patients (n = 42 per group). In total, 126 blood samples obtained from each patient were stored in EDTA and labeled as Sg1, Sg2, etc.

## *2.2. Genetic Analysis*

#### 2.2.1. DNA Extraction and Amplification and Sequencing of MT-CYB

Genomic DNA was extracted using the DNase Blood Kit (Qiagen, South Korea) in accordance with the manufacturer's instructions. Polymerase chain reaction (PCR) was carried out to amplify *MT-CYB*, since it is reportedly involved in cardiovascular pathologies [12,15–19]. PCR amplification of *MT-CYB* was carried out at a reaction volume of 50 μL containing 2 μL of concentrated DNA and 48 μL of the PCR mix comprising 29.8 μL of MilliQ water, 5 μL of buffer, 1 μL of supplementary MgCl2, 2 μL of dATP, dCTP, dGTP, and dTTP, 5 μL of H15915, 5 μL of L14723, and 0.2 μL of Tap polymerase. L14723 (5'-ACCAATGACATGAAAAATCATGGTT-3') and H15915 (5'-TCTCCATTTCTGGTTTACAAGAC-3') were the forward and reverse primers, respectively. The PCR program included the following conditions: 94 ◦C for 3 min; 40 cycles (94 ◦C for 45 s; 52 ◦C for 1 min; 72 ◦C 1 min for 30 s); 72 ◦C for 10 min. PCR products were purified and sequenced. Sequencing reactions were performed using an MJ Research PTC-225 Peltier thermocycler with the ABI PRISM kit and electrophoresed in an ABI 3730 XL sequencer.

## 2.2.2. Molecular Analyses

The chromatograms obtained after sequencing were submitted to the Mutation Surveyor software (https://softgenetics.com) version 5.0 to identify mutations and to determine their nature (homoplasmic or heteroplasmic) and their status (transition or transversion). Sequences of ARF with those of the controls. Mutation Surveyor assigned a score for each mutation, thus indicating the level of confidence regarding the accuracy of the cited base. Only those mutations with a score of ≥20 were retained (the probability that a cited base is false was 0.001; accuracy, 99%).

To determine the appropriate nucleotide position of our mutations in the mitochondrial genome, we performed BLASTn analysis (NCBI; https://ncbi.nlm.nih.gov/) with our raw control sequence. The position of each mutation and the corresponding amino acid was determined using BLASTx 2.8.0 [20], thus facilitating the identification of putative conserved domains [21].

To highlight the potential pathogenicity of non-synonymous mutations, we performed prediction analysis using three different software for transparency and reliability:


Non-synonymous *MT-CYB* mutations have been considered pathogenic if thus stipulated by at least two prediction software; if these mutations are not reported as neutral polymorphisms and if they are present in a conserved domain; heteroplasmy indicates high pathogenicity [25].

For further molecular analysis, we corrected and aligned sequences of ARF patients with those of the controls, using BioEdit version 7.1.9 [26] using the CLUSTALW algorithm [27]. Thereafter, we determined the following parameters underlying the genetic diversity of *MT-CYB* relative to each population, on the basis of which the groups were assessed and differentiated:


These parameters were determined using DNAsp 5.10.01 [28]. Other parameters elucidating the genetic diversity are the following:


These parameters were determined using Mega7 software version 7.0.26 [29], and the frequencies of amino acids was determined at the best reading frame (no stop codon).

For analysis using Mega7, it was necessary to determine the model describing the best pattern of substitution; hence, we used the following models:


#### *J. Cardiovasc. Dev. Dis.* **2019**, *6*, 36

We used the Kimura 2-parameter model [32] to determine the rate of mutations because preferential models were unavailable for this analysis. The Nei-Gojobori modified model was used to estimate substitution rates.

Thereafter, we defined the parameters of genetic diversity among populations, thus highlighting *MT-CYB* polymorphisms in ARF patients. Using the DNAsp software, we estimated the following:


Further, we evaluated genetic di fferences and structural di fferences in *MT-CYB* in accordance with the study population. Using Mega7 version 7.0.26 [29], the genetic distances were determined, which facilitated the estimation of the rates of allele replacement among the compared entities, using the JC+G model (with G = 0.36), which was the most suitable model. To estimate genetic distances within populations, we used the K2 model, which is suitable for this analysis, while the JC+G model was not. Furthermore, we estimated the genetic distance between the control population and ARF patients (operated and unoperated). For all analyses of genetic distances, the bootstrap method was used with 1000 replicates and all reading frames were accounted for. Using Arlequin version 3.1 [33], we estimated the genetic di fferentiation factor (Fst) between populations and performed analysis of molecular variance (AMOVA) to determine the origin of the variants. These analyses were performed considering polymorphic loci only and with 1023 permutations.

The Z-test for selection was performed using Mega7 in accordance with three di fferent alternative hypotheses, the null hypothesis being that H0: dN = dS. We then assessed the neutral selection (H1: dN - dS), the positive selection (dN > dS), and the negative selection (dN < dS). We used the Nei-Gojobori modified model [34], which di ffers from the original method of Nei-Gojobori in that transitional and transversional substitutions were no longer considered to occur at the same frequency. Thus, we calculated the ratio of transitions/transversions with the following formula [30]:

$$\kappa = \frac{(\pi \text{tr} \pi \text{c} + \pi \text{a} \pi \text{g})}{\pi \text{y} \pi \text{r}} \alpha / \beta \tag{1}$$

where πa, πc, πg, and πt are the nucleotide frequencies (of a, c, g, and t, respectively) estimated by their proportions in all sequences: πy = πt + πc and πr = πa + πg; α and β are the number of transitions and transversions, respectively, observed by considering all sequences in a pairwise manner.

## *2.3. Statistical Analyses*

Data normality was assessed using XLSTAT 2018.3.50896 with the Shapiro-Wilk test with a rate of significance level set to 5%. We then performed the Fisher test (for non-normally distributed data) to establish an association between mutations and the state of the disease, always at a threshold of 5%. Mean and standard deviation values for mutations were determined using MS Excel 2010 and compared using R commander (Rcmdr) implemented in RGui version 3.3.4 to assess di fferences between operated and unoperated ARF patients.

Average amino acid frequencies of the controls and unoperated AFR patients and between the controls and operated ARF patients were compared using the chi-square test in Rstudio version 1.1.447. Since chi-square analysis is only feasible when the number of groups is <5, we used complementary values (100-e ffective).
