Pharmacovariome scanning using whole pharmacogene resequencing coupled with deep computational analysis and machine learning for clinical pharmacogenomics.

BACKGROUND: This pilot study aims to identify and functionally assess pharmacovariants in whole exome sequencing data. While detection of known variants has benefited from pharmacogenomic-dedicated bioinformatics tools before, in this paper we have tested novel deep computational analysis in addition to artificial intelligence as possible approaches for functional analysis of unknown markers within less studied drug-related genes. METHODS: Pharmacovariants from 1800 drug-related genes from 100 WES data files underwent (a) deep computational analysis by eight bioinformatic algorithms (overall containing 23 tools) and (b) random forest (RF) classifier as the machine learning (ML) approach separately. ML model efficiency was calculated by internal and external cross-validation during recursive feature elimination. Protein modelling was also performed for predicted highly damaging variants with lower frequencies. Genotype-phenotype correlations were implemented for top selected variants in terms of highest possibility of being damaging. RESULTS: Five deleterious pharmacovariants in the RYR1, POLG, ANXA11, CCNH, and CDH23 genes identified in step (a) and subsequent analysis displayed high impact on drug-related phenotypes. Also, the utilization of recursive feature elimination achieved a subset of 175 malfunction pharmacovariants in 135 drug-related genes that were used by the RF model with fivefold internal cross-validation, resulting in an area under the curve of 0.9736842 with an average accuracy of 0.9818 (95% CI: 0.89, 0.99) on predicting whether a carrying individuals will develop adverse drug reactions or not. However, the external cross-validation of the same model indicated a possible false positive result when dealing with a low number of observations, as only 60 important variants in 49 genes were displayed, giving an AUC of 0.5384848 with an average accuracy of 0.9512 (95% CI: 0.83, 0.99). CONCLUSION: While there are some technologies for functionally assess not-interpreted pharmacovariants, there is still an essential need for the development of tools, methods, and algorithms which are able to provide a functional prediction for every single pharmacovariant in both large-scale datasets and small cohorts. Our approaches may bring new insights for choosing the right computational assessment algorithms out of high throughput DNA sequencing data from small cohorts to be used for personalized drug therapy implementation.

Identified PAH V230A and PAH V230I mutations in a family with diverse clinical presentations.

Phenylketonuria (PKU) is a hereditary disorder caused by phenylalanine hydroxylase enzyme (PAH) defects that might cause severe brain damage. The current main treatment, dietary management, can prevent the symptoms if commenced early. However, it has side effects if used for a long time. Additionally, some patients with mild hyperphenylalaninemia (mHPA), who has serum phenylalanine levels <360 µmol/L, do not require treatment. Since the correlation between genotype and metabolic phenotype has been demonstrated earlier, genotype-based detection of patients who do not need treatment might help with genetic counseling and choosing the most appropriate treatment option. In this study, we report an asymptomatic adult with mHPA who had never taken any medical intervention to control or lower her serum phenylalanine level (Phe). She had 179 µmol/L serum phenylalanine level and carried p.[V230A];[V230I] genotype. Her child was affected with phenylketonuria and had p.[V230A];[V230A] genotype. Both pathogenic variants detected in the asymptomatic adult with mHPA were computationally analyzed to assess their pathogenicity and the p.V230I pathogenic variant was demonstrated to be responsible for the mHPA phenotype in the asymptomatic adult detected in this study. The findings in this study could contribute to genetic counseling and treatment for families and individuals with p.[V2030I];[V230A] genotype.

Tetrahydrobiopterin responsiveness in Phenylalanine hydroxylase deficient patients from North-east of Iran: Genotype-phenotype correlation, identification of a novel mutation and 7 new responsive genotypes.

Phenylalanine hydroxylase enzyme defects result in a hereditary metabolic disorder called phenylketonuria. Sapropterin (tetrahydrobiopterin) is one of the treatment strategies for this disorder. Even though a correlation between genotype and BH4 responsiveness was established by earlier studies, a subset of mutations often presented inconsistent responses and/or phenotypes. Different genetic background is one of the potential reasons for this fact. In this study, the genotype of a total of 34 PAH deficient patients from Khorasan-Razavi providence in the north-east of Iran was obtained. Among this patients, 21 individuals took the 24 h and 48 h BH4 loading test and if the result was positive, their Phenylalanine tolerance was assessed. It is the first study of its type in patients from Iran to evaluate genotype role in predicting the most probable responsive individuals. The known pathogenic variant p.R169P and the novel variant p. Leu72_Asp75delinsTyr were first classified as responsive.Seven genotypes were reported as responsive for the first time. All patients carrying at least one pathogenic variant, which was previously reported as BH4 responsive, respond to BH4. Three patients with p.L48S, p.R261Q and p.A309V pathogenic variants were exceptions. There was no certain statistical correlation between genotype and response. Genotype and phenotype were significantly correlated and majority of patients with mild phenotype carried at least one non-null pathogenic variant. In Khorasan-Razavi province of Iran, patients with at least one non-null mutation are most probable to demonstrate mild phenotype and respond to BH4 phenotype.