Genetic and Bioinformatic Strategies to Improve Diagnosis in Three Inherited Bleeding Disorders in Bogotá, Colombia

Inherited bleeding disorders (IBDs) are the most frequent congenital diseases in the Colombian population; three of them are hemophilia A (HA), hemophilia B (HB), and von Willebrand Disease (VWD). Currently, diagnosis relies on multiple clinical laboratory assays to assign a phenotype. Due to the lack of accessibility to these tests, patients can receive an incomplete diagnosis. In these cases, genetic studies reinforce the clinical diagnosis. The present study characterized the molecular genetic basis of 11 HA, three HB, and five VWD patients by sequencing the F8, F9, or the VWF gene. Twelve variations were found in HA patients, four in HB patients, and 19 in WVD patients. From these variations a total of 25 novel variations were found. Disease-causing variations were used as positive controls for validation of the high-resolution melting (HRM) variant-scanning technique. This approach is a low-cost genetic diagnostic method proposed to be incorporated in developing countries. For the data analysis, we developed an accessible open-source code in Python that improves HRM data analysis with better sensitivity of 95% and without bias when using different HRM equipment and software. Analysis of amplicons with a length greater than 300 bp can be performed by implementing an analysis by denaturation domains.


Introduction
Inherited bleeding disorders (IBDs) are a group of conditions where deficiencies in plasma proteins, involved in blood coagulation, lead to increased risk of bleeding [1]. People suffering from bleeding disorders may experience spontaneous or increased tendency to bleed, particularly around invasive procedures. However, bleeding symptoms vary according to the severity of the disease [2]. Hemophilia A (HA), hemophilia B (HB), and von Willebrand Disease (VWD) are the most common IBDs associated with a lack of clotting factors (blood coagulation factor VIII (F8), factor IX (F9), and von Willebrand factor (VWF), respectively). Although they are rare diseases, their correct diagnosis is an important component of clinical management as it may inform treatment decisions and carrier status [3,4].
Diagnosis is usually performed through factor assays specific for coagulation proteins and genetic testing where available. However, unclear diagnosis in many cases may occur, as test results suffer from extreme variability, in part because of the wide range of methods, reagents, and instruments, and the lack of access to genetic testing [5]. This is evidenced by the differences in reports of prevalence. Data sourced from the World Bank Group show that 2639 individuals are diagnosed with HA, 591 with HB, and 2770 with VWD in Colombia. However, according to the report from the high-cost account published in The study was approved by the Ethics Committee of Los Andes University, and the methods were applied following the approved guidelines. Written informed consent was obtained from all participants. Seven patients and four carriers with HA, one patient and two carriers with HB, and five patients with VWD were selected from 20 unrelated families. Recruited patients attended the Colombian League of Hemophiliacs and Other Blood Deficiencies in Bogotá, Colombia. All patients included in the study presented a supported medical history accompanied by hematological diagnoses. A retrospective analysis of the clinical history of each family member was performed to evaluate each coagulopathy score. The following laboratory tests were evaluated: thrombin, prothrombin and thromboplastin time, ristocetin-cofactor, FVIII, VWF, and FIX9 activity, respectively.
Inclusion criteria: Colombian male and female patients of legal age with a confirmed clinical diagnosis of HA, HB, or VWD. Some hemophilia patients were underage; in these cases, blood samples were extracted from mothers as obligate carriers to avoid health complications. Carriers are all women with at least one hemophilic male in their maternal pedigree. Normal FVIII and FIX plasma levels did not rule out the carrier status of a female since FVIII or FIX levels show considerable variability, and normal levels do not always represent non-carrier status. The phenotype characteristics of patients are listed in Tables S1-S5.
Exclusion criteria: Patients with a diagnosis of acquired hemophilia A or any other bleeding disorder or unknown hemostatic defects; patients not of legal age.
Eight healthy males and eight healthy females were included as controls for HRM results validation.
Inclusion criteria: Colombian male and female subjects of legal age. Exclusion criteria: Subjects with a diagnosis of acquired hemophilia A or any other bleeding disorder or unknown hemostatic defects; patients not of legal age.

DNA Extraction
Blood samples were taken at the Haemostasis Reference Laboratory in Bogotá, Colombia. Samples were transported in the first 4 h to the Human Genetics Laboratory of Los Andes University. DNA was extracted from peripheral blood leukocytes using the proteinase K and phenol-chloroform extraction method [18]. The sample concentration was determined by NanoDrop™.

Genetic Analysis of F8, F9, and VWF Genes
In patients with HA diagnosis, the presence of intron 22 and intron 1 inversion was evaluated. Intron 22 inversion detection was performed according to the protocol followed by Polanía et al. [19]. Intron 1 inversion was evaluated through the PCR protocol described by Bagnall et al. (2012) [20]. The PCR-multiplex temperature conditions for inversion 1 included an initial denaturation at 95 • C for 2 min, followed by 30 cycles composed of denaturation at 94 • C for 30 s, alignment at 63 • C for 45 s, and an extension at 72 • C for 2 min. Subsequently, a 1% agarose gel was run at 70 V for 60 min to observe and analyze the expected products. TrackIt 1 Kb plus DNA ladder (Invitrogen, USA was used. To detect gene variations, forward and reverse primers were designed for the VWF and F9 genes (F9 ID: 2158 and VWF ID: 7450) (shown in Table 1) based on the NCBI reported sequence. The possible variations present in VWF exons 17-25 (VWD type 2N) and 28 (VWD types 2A, 2B, and 2M) were genetically evaluated to confirm or rule out VWD type 2 and type 1 diagnosis [21]. WVD type 3 was discarded considering the clinic of patients. All primers were designed and validated with the help of the NCBI tool [22], and the Genome Browser of Santa Cruz University of California [23]. F8 primer sequences were obtained from the Hadb database [19]. Samples were amplified under the following PCR conditions: 95 • C for 2 min, followed by 30 cycles (that varied according to each primer pair, see Table 1) that, for the case of F8 primers, included 94 • C for 30 s, 59 • C for 45 s, and 72 • C for 90 s. The final extension was 72 • C for 5 min. Finally, electrophoresis was performed to evaluate the expected sizes for each amplicon in a 2% agarose gel at 70 V for 60 min. Subsequently, the PCR products were sequenced by the Sanger method in the ABI prism 3500 ® genetic analyzer.
Among the criteria used to analyze the results obtained from the different predictors are the following: major criteria: finding null variants; strong criteria: amino acid change previously established as a pathogenic variant, in vitro or in vivo established pathogenic variant, functional studies supporting a deleterious effect on the gene product, prevalence of the variant in affected individuals compared to controls or evidence to be pathogenic; moderate criteria: located in a mutational and/or critical hotspot point and well-established functional domain, absence of the variant in control groups, for recessive disorders that are detected in trans with a pathogenic variant, changes in protein length as a result of in-frame deletions or insertions in a region without repeats, amino acid changes determined to be pathogenic, moderate pathogenicity, de novo variant but without confirmation of paternity or maternity; supporting criteria: multiple bioinformatic predictors support a detrimental effect, the patient's phenotype or family history is highly specific for a disease with a single genetic etiology, when a reliable source reports it as a pathogen without there being any functional studies.

High-Resolution Melting Analysis
After fully screening each patient's genotype with routine methods (Sanger sequencing), exons with variations were used as positive controls for the HRM technique validation. Analyses were performed under the Roche protocol for variation detection (LightCycler ® 96 System User Training Guide, v2.0) in a real-time PCR LightCycler ® 96 System. Melt-Doctor ™ HMR (Thermo Fisher) reagents were used, along with the respective primers for each reaction, DNA template (25 ng/µL), and water. After real-time PCR, a dissociation of the amplification was performed with a fusion step. The thermal profile corresponds to a preincubation step of 95 • C for 10 min with a ramp of 4.4 • C/s. This step was followed by a 3-step amplification that included: denaturation at 95 • C for 15 s, hybridization at a specific temperature according to each exon for 20 s, and amplification at 72 • C for 20 s, for 45 cycles. This was followed by a high-resolution melting step at 95 • C for 60 s, 40 • C for 60 s with a ramp of 2.2 • C/s, 65 • C for 1 s, and 97 • C for 1 s with a ramp 0.04 • C/s with maximum 25 readings/ • C corresponding to the highest number of fluorescence readings per degree.
For male DNA analysis, hemizygous DNA was mixed with healthy male DNA. This mixture improves the detection performance, as previously reported [8,29,30].

High-Resolution Melting Data Analysis
Values obtained from the DNA fusion curves were analyzed with the High-Resolution Melt software for LightCycler 96 System Software and, to avoid manual analysis of the HRM melting curves, we developed an algorithm in Python based on the algorithm proposed by Li et al. (2016) [30]. We extended the algorithm to allow for domain identification and analysis and analysis against multiple control curves considering the possible deviations of normal curves. Instead of using predefined angles to detect the start and end of the melting region, we used the second derivative of the melting curve to detect minimum points and peaks. The code is available at: https://gitfront.io/r/user-5547184/d3ec106 2bc7d0d2e3039b74770330c06f510d65e/hrm-analysis/, accessed on 20 August 2021.

Domain Identification
Melting large amplicons, such as exons with a size greater than 300 bp, can result in multiple melting domains, which complicate the HRM analysis. Usually, the solution for this is to create smaller amplicons, but this can increase the cost of the analysis as multiple samples are needed. Instead, we analyzed each domain separately. A domain is identified as a peak in the derivative curve. A peak in a curve is detected when its derivative changes from negative to positive. To identify domains in the melting curve, we applied a Savitsky-Golay filter using the savgol_filter function of the scipy signal library for Python to identify the second derivative of the curve. We then identified the points where it changes its sign from positive to negative. These are considered the peaks in the first derivative of the melting curve. Peaks with values below a threshold were not considered a domain for the analysis, as they were found to be noise. The threshold value was set to 0.10 by analyzing the number of domains in most curves. Each of the resulting peaks represented a domain in the melting curve. For each domain, we then identified its melting region.

Domain Melting Region Identification and Analysis
To identify the melting region of each domain, we found the minimum points in the second derivative. These are identified as the points where the sign changes from positive to negative. The start of the melting region was set 2 degrees before the start point of the domain and the end of the melting region as 2 degrees after the end of the domain, as suggested by Li et al. (2016) [30]. For each domain, curve normalization and background subtraction were performed following the procedure suggested by Li et al (2016) [30]. The melting temperature corresponded with the peak of the domain found previously.

Difference Plot
The preferred data display for high-resolution amplicon melting analysis is a difference plot which shows the difference between the analyzed curve and the mean curve of the set of controls. One difference plot for each domain was created. The median was chosen instead of the mean curve because the median curve is less affected by outliers. We also plotted the standard deviation of the median curve (shown as a gray area) to represent the variability of the set of controls. Table 2 reports the main variations found in the study sample associated with the diagnosis of each patient as well as the effect of each variation in the F8, F9, or VWF protein according to the results obtained by the different predictors. Inversion 1 ( Figure 1) and inversion 22 diagnosis was ruled out in all HA patients. Twelve variations were found in HA patients, four in HB patients and 19 in WVD patients. From these variations, a total of twenty-five novel variations were found. Table 3 summarizes and classifies the variations found in the population under study, according to their effect on the protein and their report in the literature. To our knowledge, these new variants have not been reported in other previous studies carried out in the Colombian population [10].

Development of an Open-Source Code in Python for HRM Data Analysis
HRM analysis was enriched through an open-source code in Python. Steps included the following: to extract each domain, the melting curve was preprocessed using the Savistzky-Golay filter to obtain the first derivative curve of each sample. Using this derivative curve, all peaks were identified by finding the point where the second derivative changes from negative to positive (see Figure 2A which shows four peaks). Peaks with values below a threshold were not considered a domain for the analysis, as they were found to be noise. The threshold value was set to 0.10 by analyzing the number of domains in most curves. Each of the resulting peaks represented a domain in the melting curve. For each domain, the start and end of the melting region were found by finding the minimum points of the curve that enclosed the peak. To do this, the points found in the second derivative where the value changes from positive to negative (minimum points of the curve). A pseudo-algorithm of this process is shown in Figure 2B.
After finding the domains, each of them was analyzed as a new melting region following the steps in the procedure by Li et al. (2016) [30]. The start of the melting region was set 2 degrees before the start point of the domain and the end of the melting region as 2 degrees after the end of the domain, as suggested by the authors. For each domain, curve normalization and background subtraction were performed, as well as finding the melting temperature. The melting temperature corresponded with the peak of the domain found previously.
Finally, for the difference plot construction, one difference plot for each domain was created, comparing with the median curve for the same domain of a set of control experiments. Each melting curve in the dataset was subtracted from the median curve. The median was chosen instead of the mean curve as Li et al. (2016) [30] did because the median curve is less affected by outliers.

High-Resolution Melting Technique Validation
In a preliminary way, we wanted to evaluate the pertinence of the HRM technique in identifying point variations. For this purpose, a point variation (c.440T > C) found on exon 4 of the F8 in two HA patients was selected as a positive control. Data were analyzed with the commercial software from Roche. As shown in Figure 3, it is evident that despite the variation of the data, the average behavior of the wildtype and mutant curves is different. Differences were judged significant if the curves of a mutated amplicon were found with similar values outside the range of normality.         and inversion 22 diagnosis was ruled out in all HA patients. Twelve variations were found in HA patients, four in HB patients and 19 in WVD patients. From these variations, a total of twenty-five novel variations were found. Table 3 summarizes and classifies the variations found in the population under study, according to their effect on the protein and their report in the literature. To our knowledge, these new variants have not been reported in other previous studies carried out in the Colombian population [10].   For each domain, the start and end of the melting region were found by finding the minimum points of the curve that enclosed the peak. To do this, the points found in the second derivative where the value changes from positive to negative (minimum points of the curve). A pseudo-algorithm of this process is shown in Figure 2B. After finding the domains, each of them was analyzed as a new melting region following the steps in the procedure by Li et al. (2016) [30]. The start of the melting region was set 2 degrees before the start point of the domain and the end of the melting region as 2 degrees after the end of the domain, as suggested by the authors. For each domain, curve normalization and background subtraction were performed, as well as finding the melting temperature. The melting temperature corresponded with the peak of the domain found previously.
Finally, for the difference plot construction, one difference plot for each domain was created, comparing with the median curve for the same domain of a set of control experiments. Each melting curve in the dataset was subtracted from the median curve. The median was chosen instead of the mean curve as Li et al. (2016) [30] did because the median curve is less affected by outliers. This preliminary validation with the point variation was extended to the other variations previously found by Sanger sequencing from the other patients with HA and HB to confirm the effectiveness of the technique in scanning variations. To improve our data analysis, denaturation curves were analyzed using Python programming code. By scanning variations in the exons of patients diagnosed with HA or HB, curve profiles other than the normal type were found. Table 4 summarizes HRM data results, where 21 out of a total of 22 variations were validated. Figure 4A-C show the HRM analysis of variation c.440T > C on exon 4 of the F8 with the Python code. Better differentiation of the samples can be evidenced for healthy controls in contrast to the graphs obtained with the commercial software shown in Figure 3.
For exons with a size less than 300bp, the analysis by HRM using the code was successful. For example, Figure 4D-F show the HRM analysis for exon 7, that has 150 bp, of sample Af1B. This sample presented a nucleotide insertion and shows a clearly differentiable curve concerning samples from healthy individuals. Analysis of the variants found in samples af2A, Am2a, Am3a, Am5a, Am6a from exon 24 was also successful ( Figures S2 and S3), however, in these samples, it was necessary to discriminate the analysis by sex to obtain better results. It was not possible to identify a different curve profile for exon 6 from sample Af2A, because the curve of the negative derivative plot was within the variation pattern of the curves from healthy controls, probably because of the position of the variation in the sequence.

High-Resolution Melting Technique Validation
In a preliminary way, we wanted to evaluate the pertinence of the HRM technique in identifying point variations. For this purpose, a point variation (c.440T > C) found on exon 4 of the F8 in two HA patients was selected as a positive control. Data were analyzed with the commercial software from Roche. As shown in Figure 3, it is evident that despite the variation of the data, the average behavior of the wildtype and mutant curves is different. Differences were judged significant if the curves of a mutated amplicon were found with similar values outside the range of normality. This preliminary validation with the point variation was extended to the other variations previously found by Sanger sequencing from the other patients with HA and HB to confirm the effectiveness of the technique in scanning variations. To improve our data analysis, denaturation curves were analyzed using Python programming code. By scanning variations in the exons of patients diagnosed with HA or HB, curve profiles other than the normal type were found. Table 4 summarizes HRM data results, where 21 out of a total of 22 variations were validated. Figure 4A-C show the HRM analysis of variation c.440T > C on exon 4 of the F8 with the Python code. Better differentiation of the samples can be evidenced for healthy controls in contrast to the graphs obtained with the commercial software shown in Figure 3.  tiable curve concerning samples from healthy individuals. Analysis of the variants found in samples af2A, Am2a, Am3a, Am5a, Am6a from exon 24 was also successful ( Figures S2  and S3), however, in these samples, it was necessary to discriminate the analysis by sex to obtain better results. It was not possible to identify a different curve profile for exon 6 from sample Af2A, because the curve of the negative derivative plot was within the variation pattern of the curves from healthy controls, probably because of the position of the variation in the sequence. Exons with a length greater than 300 bp presented melting curves with several melting domains. To improve sensitivity, analysis by denaturation domains was implemented [13]. The analysis was performed in two steps, first studying the overall amplicon and then each domain. Figure 5 shows the domain analysis for a point variation (g.4266C > T) on exon 1 of the F8 found in one HA patient. For this experiment, three healthy subjects were used as controls for validations. Figure 5 shows the identification of each domain within a sample, while Figure 6 shows the HRM analysis by domain. In this experiment, Figure 6 shows the healthy controls in gray and the shadow shows the mean with its deviation. The analysis (shown in yellow) corresponds to the positive control, where we have the DNA of the patient that corresponds to an affected male and the same patient as an artificial heterozygote. In Figure 6B, the three graphs (melt curve, negative derivative plot, and difference plot) obtained for the analysis of domain 2 are shown, and in this case, we see that in this domain there are no differences between healthy controls and positive controls. Therefore, in that region of the amplified fragment, we can deduce that there are no differences in the DNA sequence. On the other hand, domains 1 and 3 show us differences in both the patient's DNA and the artificial heterozygote of the same patient. In this case, when finding these differences in a blind test or with an unknown sample, that exon of that patient should be sequenced to identify the specific variation. Figure 6C shows the greater differences between samples and healthy controls. As seen in Figure 5B, in all the samples analyzed in this experiment, domain 3 was detected, and as evidenced in Figure 6C, it is the most informative domain to detect differences in this experiment. Finally, we propose a protocol for molecular analysis using HRM that supports the clinical diagnosis of patients with hemophilia, shown in Figure 7.
case, when finding these differences in a blind test or with an unknown sample, that exon of that patient should be sequenced to identify the specific variation. Figure 6C shows the greater differences between samples and healthy controls. As seen in Figure 5B, in all the samples analyzed in this experiment, domain 3 was detected, and as evidenced in Figure  6C, it is the most informative domain to detect differences in this experiment. Finally, we propose a protocol for molecular analysis using HRM that supports the clinical diagnosis of patients with hemophilia, shown in Figure 7.      Proposed protocol for genetic analysis using HRM. In this protocol, we propose to take the DNA samples to carry out the HRM test and, in the case of men, an artificial heterozygote should be included in the test run. In the case of exons that are reported as highly polymorphic, it is recommended to sequence without a previous scan by HRM, as they will probably give positive results. Once the data from the qPCR equipment are obtained at the time of the analysis, we recommend discriminating the samples by sex, and in the case of exons with a size greater than 300 bp, to include an analysis by domains. Samples that are positive through these analyses must be confirmed by Sanger sequencing.

Genetic Analysis of F8 Gene
Genetic analyses in all cases were performed under the HGVS Recommendations for the Description of Sequence Variants [31]. The intron 22 and 1 inversions (Inv22 and Inv1) are the most frequent molecular alterations found in severe HA patients with a frequency of 45-50% and 0.5-5%, respectively. However, in a study carried out in Colombia by Yunis et al. (2018), inversion 22 was detected in 14/33 male patients (42.4%), and inversion 1 was detected in 3/33 [10]. These results are consistent with our study since it must be considered that, in total, we had 11 patients with hemophilia A, from which four were men classified as severe, three men classified as mild, and four carriers. In this scenario, it is consistent to find negative results of inversion assays.
Patient Af1A is an HA carrier with low F8 levels (24.3%) (Tables S1 and S2). This patient presents a c.3690_3691insG in exon 14 ( Figure 8). The variation results in a truncated protein, so it should have clear functional consequences. Patient Af2A is also an HA carrier, despite having normal FVIII levels. This phenomenon has been observed in other studies where bleeding tendency is also observed in some hemophilia A carriers with normal factor VIII levels [32]. This patient presents variations c.673A > G and c.6605A > T. Supporting criteria for these variations include multiple bioinformatic predictors that sup- Proposed protocol for genetic analysis using HRM. In this protocol, we propose to take the DNA samples to carry out the HRM test and, in the case of men, an artificial heterozygote should be included in the test run. In the case of exons that are reported as highly polymorphic, it is recommended to sequence without a previous scan by HRM, as they will probably give positive results. Once the data from the qPCR equipment are obtained at the time of the analysis, we recommend discriminating the samples by sex, and in the case of exons with a size greater than 300 bp, to include an analysis by domains. Samples that are positive through these analyses must be confirmed by Sanger sequencing.

Genetic Analysis of F8 Gene
Genetic analyses in all cases were performed under the HGVS Recommendations for the Description of Sequence Variants [31]. The intron 22 and 1 inversions (Inv22 and Inv1) are the most frequent molecular alterations found in severe HA patients with a frequency of 45-50% and 0.5-5%, respectively. However, in a study carried out in Colombia by Yunis et al. (2018), inversion 22 was detected in 14/33 male patients (42.4%), and inversion 1 was detected in 3/33 [10]. These results are consistent with our study since it must be considered that, in total, we had 11 patients with hemophilia A, from which four were men classified as severe, three men classified as mild, and four carriers. In this scenario, it is consistent to find negative results of inversion assays.
Patient Af1A is an HA carrier with low F8 levels (24.3%) (Tables S1 and S2). This patient presents a c.3690_3691insG in exon 14 ( Figure 8). The variation results in a truncated protein, so it should have clear functional consequences. Patient Af2A is also an HA carrier, despite having normal FVIII levels. This phenomenon has been observed in other studies where bleeding tendency is also observed in some hemophilia A carriers with normal factor VIII levels [32]. This patient presents variations c.673A > G and c.6605A > T. Supporting criteria for these variations include multiple bioinformatic predictors that support a detrimental effect and the patient's family history is highly specific for the disease.
Variation c.6605A > T was also found in patient Am6A whose phenotype is mild (Tables S1 and S2). In these two cases, both patients present the same variation, however, each variation's phenotypic penetrance and expressivity vary due to the different combinations of modifying alleles that are present in one genetic background versus another. In addition, the penetrance of some pathogenic genotypes is known to be age and/or sex dependent. Variable penetrance may also reflect the action of unlinked modifier genes, epigenetic changes, or environmental factors. Some examples of pathogenic microlesions whose penetrance has been found to be modulated by allelic SNPs are provided by Cooper et al. [33]. In the study reported by Venceslá et al. (2010), three consanguineous sisters presented the same homozygous variation. However, two sisters had moderate bleeding due to a known mutation, and one of the sisters had no bleeding history [34]. Patient Am1A, with a mild phenotype, presented variation c.5375T > A which was also predicted as pathogenic. Patient Am2A is a severe HA case. Variation c.203C > T is reported by the HGMD (CM105565) as a known disease variation (Figure 8). This variation was also found in a severe HA patient in Sweden [35]. This patient has a positive inhibitor test result, and it is known that patients who develop inhibitors usually have significant changes in the F8 sequence [36]. Patients Af3A and Af4A are both HA carriers with variation c.440T > C. Both patients have sons with severe phenotypes. This variation is a known disease variation reported by the HGMD (CM021585) (Figure 8). Patient Af3A is a symptomatic carrier with bleeding tendency. This phenomenon has also been observed in some hemophilia A carriers with normal factor VIII levels as in this case and requires further investigation (Tables S1 and S2) [32,[37][38][39]. Shinozawa et al. (2021) mention the impact of age and blood group on FVIII:C level [4]. Biguzzi et al. (2021) reported that VWF:Ag and FVIII:C increase with age [40]. Carriers of a non-O blood group present a steeper increase in VWF:Ag and FVIII:C with age, which is mediated by acquired factors. Additionally, it is consistent with Af3A's blood type B and her age. Many studies have reported that there are great inter-and intraindividual variations in coagulation markers in women due to different physiological conditions such as age, ethnicity, blood group, and phases of the menstrual cycle [41]. Other studies, such as the one reported by Miesbach et al. (2011), show that female carriers of hemophilia A can present with FVIII:C levels within the normal range but might also report a considerable tendency to bleed [42]. Even carriers with normal FVIII:C activity present an increased risk of bleeding as does The only way to ascertain the carrier status is through a molecular diagnosis of the causative mutation in F8 or F9 as we did. Gene analysis is the gold standard for carrier diagnosis as it identifies undetected female carriers. Despite the clear advantage of nextgeneration sequencing in various settings, Sanger sequencing remains a better option for genetic testing and diagnosis of hemophilia carriers. There may still be a variety of unknown mutations of F8 and F9 target genes as our research and Shinozawa et al. (2021) reported [4]. Any new variation found by NGS must still be validated by Sanger sequencing. It is clearly seen that corroborating the carrier status of women, whose factor measurements may be normal, is necessary through genetic analysis.
Variation c.6605A > T was also found in patient Am6A whose phenotype is mild (Tables S1 and S2). In these two cases, both patients present the same variation, however, each variation's phenotypic penetrance and expressivity vary due to the different combinations of modifying alleles that are present in one genetic background versus another. In addition, the penetrance of some pathogenic genotypes is known to be age and/or sex dependent. Variable penetrance may also reflect the action of unlinked modifier genes, epigenetic changes, or environmental factors. Some examples of pathogenic microlesions whose penetrance has been found to be modulated by allelic SNPs are provided by Cooper et al. [33]. In the study reported by Venceslá et al. (2010), three consanguineous sisters presented the same homozygous variation. However, two sisters had moderate bleeding due to a known mutation, and one of the sisters had no bleeding history [34].
Patient Am1A, with a mild phenotype, presented variation c.5375T > A which was also predicted as pathogenic. Patient Am2A is a severe HA case. Variation c.203C > T is reported by the HGMD (CM105565) as a known disease variation (Figure 8). This variation was also found in a severe HA patient in Sweden [35]. This patient has a positive inhibitor test result, and it is known that patients who develop inhibitors usually have significant changes in the F8 sequence [36]. Patients Af3A and Af4A are both HA carriers with variation c.440T > C. Both patients have sons with severe phenotypes. This variation is a known disease variation reported by the HGMD (CM021585) (Figure 8). Patient Af3A is a symptomatic carrier with bleeding tendency. This phenomenon has also been observed in some hemophilia A carriers with normal factor VIII levels as in this case and requires further investigation (Tables S1 and S2) [32,[37][38][39]. Shinozawa et al. (2021) mention the impact of age and blood group on FVIII:C level [4]. Biguzzi et al. (2021) reported that VWF:Ag and FVIII:C increase with age [40]. Carriers of a non-O blood group present a steeper increase in VWF:Ag and FVIII:C with age, which is mediated by acquired factors. Additionally, it is consistent with Af3A's blood type B and her age. Many studies have reported that there are great inter-and intraindividual variations in coagulation markers in women due to different physiological conditions such as age, ethnicity, blood group, and phases of the menstrual cycle [41]. Other studies, such as the one reported by Miesbach et al. (2011), show that female carriers of hemophilia A can present with FVIII:C levels within the normal range but might also report a considerable tendency to bleed [42]. Even carriers with normal FVIII:C activity present an increased risk of bleeding as does this female patient. Incidence and intensity of bleeding symptoms of hemophilia A carriers are high and correlated with the phenotype of the male hemophilic relative and the underlying F8 gene mutation as in this case.
Patient Am3A has a severe HA diagnosis (F8 levels were at 1.3%) (Tables S1 and S2). This patient has variation c.789T > C that is altered within the used splice site, likely to disturb normal splicing, affecting protein features. Patients Am4A and Am7A both have severe phenotypes (Tables S1 and S2). Patient Am4A presents variation c.5506T > G and patient Am7A presents variation c.673A > G. Finally, patient Am5A has an unknown classification of his phenotype. His brothers have a confirmed diagnosis of HA (Tables S1 and S2); This patient presents variation c.6605A > T. This patient may have a higher risk of bleeding or a mild phenotype. For all cases, it is recommended to perform expression studies and functional analysis of the new variations found to demonstrate that these sequence variations affect F8 synthesis or function, as they are classified as variants with uncertain significance.

Genetic Analysis of F9 Gene
Patient Af1B is a possible carrier, without classification of disease severity (Table S3). The two variations found produce a protein strongly truncated with the loss of the peptidase S1 domain. This prevents the formation of a buried saline bridge [43]. Patient Am1B presents a severe HB phenotype with a family history of the disease in mother and siblings. The pathogenic variant rs137852261 found in this patient is associated with hereditary F9 deficiency (ID HGMD: CM940671). As reported by Ludwig et al. (1989) [44], a transition from C to T in the non-coding strand may explain these individual nucleotide substitutions. CpG dinucleotides are variation hot spots due to cytosine methylation followed by spontaneous deamination to thymine. No immune response related to the production of inhibitors has been reported in this patient as in the study [44,45]. Finally, in patient Af2B, a deletion of a single nucleotide was found. This deletion is predicted to produce a change in the reading frame or premature termination codons (PTCs), producing a strongly truncated protein, which may cause a "nonsense-mediated decay" (NMD). These results match the supporting criteria that include multiple bioinformatic predictors that support a detrimental effect and the patient's phenotype, or family history is highly specific for a disease with a single genetic etiology.
The only high-frequency polymorphism in the F9 is an A to G transition known as the Malmö G/A dimorphism mapped to exon 6 [8]. This single nucleotide polymorphism (ID rs6048) has a G allele frequency of 20%, however, none of our patients presented the polymorphism.

Genetic Analysis of VWF Gene
For the VWD patient analysis, missense variations clustered in exon 28 usually associate with VWD types 2A, 2M, or 2B [21]. Patient Af1VW presents variations c.3835G > A and c.4133C > T. This patient is a compound heterozygote for defects in the VWF as found in the Eikenboom et al. study (1993) [46]. Since the variation is within the domains of the VWF that are involved in binding to the platelet membrane glycoprotein Ib as shown in Figure 9. The mutations may interfere with the ristocetin-induced VWF binding to platelets. This variation was found in patients with VWD type I diagnosis [46]. Variation c.4133C > T was found in a cohort study, where it was identified as type 1 VWD [47]. A broad range of different mutant VWF allele types were present in the study population and were inherited alone or in combination, resulting in a complex array of VWD types as in this case. Both variations correlate with low VWF levels (42.9) in this patient and these results coincide with the strong correlation criteria previously defined in the methodology section. Patient Af2VW presents one non-synonymous substitution, Y1584C. This variation is a known disease variation (HGM CM031758). rs1800386 has been associated with the type 1 VWD phenotype as reported in previous studies [48][49][50][51]. As a future perspective, it is important to evaluate the multimer pattern to discard a type 2A VWD and confirm that the variation behaves as in the European and North American population as the cause of a type 1 VWD [48][49][50][51]. Patient Af3VW presents four variations in exons 19 and 25 (c.2479T > A, c.2482C > A, c.3291C > G, and c.3350T > G). It has been reported that variations in VWF that specifically decrease binding to F8 (type 2N VWD) map to both the TIL' and E' segments, suggesting a direct role in binding factor VIII [52]. Nevertheless, F8 levels in this patient are normal (Tables S4 and S5) so a VWD type 2N diagnosis must be carefully given. In VWD type 1, gene defects are spread throughout the entire VWF [53], so these variations found may be associated with a VWD type 1 phenotype.
For patient Am1VW, there was a clinical diagnosis of VWD type 2N (Tables S4 and  S5). Variation c.2535C > A in exon 19 was found. The variation is associated with a loss of the VWF D3 domain ( Figure 5). The variation found and a previous diagnosis of HA could correlate to a VWD type 2N diagnosis in this patient. Patient Af4VW presents the following variations: c.3252C > G, c.3297C > G, and c.3350T > G in exon 25 ( Figure 5). It is important to mention that none of the VWD patients included in the study had been diagnosed with the multimer analysis, confirming the importance of complementing the clinical diagnosis with a genetic diagnosis. According to these protein effects, this patient might be a case of VWD type A2 with enhanced susceptibility to ADAMTS13-mediated proteolysis, impaired dimer and multimer assembly, loss of regulated storage, and intracellular retention. However, the different effects might also be associated with types 2N or 2M. A multimer analysis must be performed to fully diagnose this patient. For all cases, it is recommended to perform expression studies and functional analysis of the new variations found to demonstrate that these sequence variations affect VWF synthesis or function.

High-Resolution Melting Technique Validation
Two different patients with the same variation, c.440T > C, were tested and the same Patient Af2VW presents one non-synonymous substitution, Y1584C. This variation is a known disease variation (HGM CM031758). rs1800386 has been associated with the type 1 VWD phenotype as reported in previous studies [48][49][50][51]. As a future perspective, it is important to evaluate the multimer pattern to discard a type 2A VWD and confirm that the variation behaves as in the European and North American population as the cause of a type 1 VWD [48][49][50][51]. Patient Af3VW presents four variations in exons 19 and 25 (c.2479T > A, c.2482C > A, c.3291C > G, and c.3350T > G). It has been reported that variations in VWF that specifically decrease binding to F8 (type 2N VWD) map to both the TIL' and E' segments, suggesting a direct role in binding factor VIII [52]. Nevertheless, F8 levels in this patient are normal (Tables S4 and S5) so a VWD type 2N diagnosis must be carefully given. In VWD type 1, gene defects are spread throughout the entire VWF [53], so these variations found may be associated with a VWD type 1 phenotype.
For patient Am1VW, there was a clinical diagnosis of VWD type 2N (Tables S4 and S5). Variation c.2535C > A in exon 19 was found. The variation is associated with a loss of the VWF D3 domain ( Figure 5). The variation found and a previous diagnosis of HA could correlate to a VWD type 2N diagnosis in this patient. Patient Af4VW presents the following variations: c.3252C > G, c.3297C > G, and c.3350T > G in exon 25 ( Figure 5). It is important to mention that none of the VWD patients included in the study had been diagnosed with the multimer analysis, confirming the importance of complementing the clinical diagnosis with a genetic diagnosis. According to these protein effects, this patient might be a case of VWD type A2 with enhanced susceptibility to ADAMTS13-mediated proteolysis, impaired dimer and multimer assembly, loss of regulated storage, and intracellular retention. However, the different effects might also be associated with types 2N or 2M. A multimer analysis must be performed to fully diagnose this patient. For all cases, it is recommended to perform expression studies and functional analysis of the new variations found to demonstrate that these sequence variations affect VWF synthesis or function.

High-Resolution Melting Technique Validation
Two different patients with the same variation, c.440T > C, were tested and the same pattern of behavior for the samples in the different plot views is evidenced. Due to the position of the variation at the beginning of the amplicon, the signal magnitude was very close to that of the reference curve as shown in Figure 3B. However, the technique is sensitive enough to show differences. Nakagawa et al. (2016) [54] performed an HRM assay to detect a C to T transition located at nucleotide position c.1423 in exon 12 of the ADAMTS13 gene and demonstrated the potential benefit of the HRM technique for genotyping purposes. In replicate, the sensitivity and specificity of HRM were found to be identical to this study [54]. Better differentiation of the samples was evidenced when analyzing data with the programming code in Python vs. the graphs obtained with the commercial software.
According to [8,29,55], a mixture of male DNA with wildtype male DNA was necessary to generate an artificial heterozygote and detect the formed duplex. There is strong evidence of the benefit of mixing shown by our study, however, we also detected in the variation of the Am7A sample for exon 3 of F8 that the difference plot revealed a greater difference of the sample without a DNA mixture with respect to normal samples than the artificial heterozygote ( Figure S1). For this reason, we suggest including both types of samples in the exon analysis for males.
It is important to take into consideration that many authors have reported that a larger size of the amplicons is associated with lower efficiency of the method [8,[11][12][13]54]. However, confirmed variations present in exons 8.1 and 8.3 of the F9 from Af2B, Am1B, and Af1B patients, and variation on exon 1 of the F8, were detected despite the size of the amplicon (800 bp). In these cases, domain analysis allowed us to improve the previously reported sensitivity of 93% [11] to 95% at a lower cost. Unlike the methodology used by other studies where 52 reactions per patient were needed to carry out the gene analysis [10,11], we only used 35 reactions per patient. We propose a protocol for molecular analysis using HRM that supports the clinical diagnosis of patients with hemophilia, shown in Figure 7.

Conclusions
Twenty-seven novel variations were found in this study. As shown in this study, sequencing of exons associated with VWD type 2 phenotypes helps to better guide VWD patient diagnosis and should be incorporated into the routine analysis.
HRM is a promising tool for the molecular diagnosis of these coagulopathies as it has proven to be a sensitive technique that can be used as a diagnostic method and should be incorporated in developing countries. Given the great socioeconomic impact of these diseases on different countries' health systems, HRM represents significant cost differences in contrast to direct gene sequencing that can be translated not only into a greater number of patients who can access an accurate diagnosis but also into an effective treatment and a better quality of life.
The use of programming code improves the objectivity and the performance of the measurements generated with the HRM analysis. Analysis discriminated by sex improves the sensitivity and, in the case of amplicons with a length greater than 300 bp, domain analysis also improves sensitivity, making the analysis more cost-effective. Among the advantages, we include the objective identification of the Tm1 and Tm2 since these temperatures represent subjective parameters when working only with the LightCycler program. Open-source codes allow continuous improvement of the analysis by applying more functions that allow a more objective analysis that is not possible in commercial programs.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/10 .3390/genes12111807/s1, Figure S1: Variation c.673A > G HRM analysis, with male discrimination. Difference plot shows greater difference of the sample without DNA mixture (shown with a red arrow) with respect to normal samples than the artificial heterozygote; Figure S2: Variation c.6605A > T HRM analysis, with female discrimination; Figure S3: F8 exon 24 HRM analysis, with male discrimination; Table S1: Phenotype characteristics of hemophilia A patients; Table S2: Phenotype characteristics of hemophilia A patients; Table S3: Phenotype characteristics of hemophilia B patients; Table S4: Phenotype characteristics of von Willebrand Disease patients; Table S5