DataSheet1_Which Is the Best In Silico Program for the Missense Variations in IDUA Gene? A Comparison of 33 Programs Plus a Conservation Score and Evaluation of 586 Missense Variants.xlsx

Mucopolysaccharidosis type I (MPS I) is an autosomal recessive disease characterized by the deficiency of alpha-L-iduronidase (IDUA), an enzyme involved in glycosaminoglycan degradation. More than 200 disease-causing variants have been reported and characterized in the IDUA gene. It also has several variants of unknown significance (VUS) and literature conflicting interpretations of pathogenicity. This study evaluated 586 variants obtained from the literature review, five population databases, in addition to dbSNP, Human Genome Mutation Database (HGMD), and ClinVar. For the variants described in the literature, two datasets were created based on the strength of the criteria. The stricter criteria subset had 108 variants with expression study, analysis of healthy controls, and/or complete gene sequence. The less stringent criteria subset had additional 52 variants found in the literature review, HGMD or ClinVar, and dbSNP with an allele frequency higher than 0.001. The other 426 variants were considered VUS. The two strength criteria datasets were used to evaluate 33 programs plus a conservation score. BayesDel (addAF and noAF), PON-P2 (genome and protein), and ClinPred algorithms showed the best sensitivity, specificity, accuracy, and kappa value for both criteria subsets. The VUS were evaluated with these five algorithms. Based on the results, 122 variants had total consensus among the five predictors, with 57 classified as predicted deleterious and 65 as predicted neutral. For variants not included in PON-P2, 88 variants were considered deleterious and 92 neutral by all other predictors. The remaining 124 did not obtain a consensus among predictors.

黏多糖贮积症I型（Mucopolysaccharidosis type I, MPS I）是一类常染色体隐性遗传病，以α-L-艾杜糖醛酸酶(alpha-L-iduronidase, IDUA)缺乏为特征——该酶参与糖胺聚糖的降解代谢。目前已在IDUA基因中鉴定出200余种致病变异，同时存在大量意义未明变异(variants of unknown significance, VUS)，且相关文献对其致病性的解读存在分歧。本研究对通过文献综述、5个人群数据库，以及dbSNP、人类基因突变数据库(Human Genome Mutation Database, HGMD)及ClinVar获取的586个变异开展了评估。针对文献报道的变异，本研究依据评估标准的严格程度构建了两套数据集：严格标准子集包含108个变异，均带有表达研究、健康对照分析及/或完整基因序列数据；宽松标准子集则额外纳入了52个通过文献综述、HGMD或ClinVar及dbSNP获取、且等位基因频率高于0.001的变异。剩余426个变异被归类为意义未明变异。本研究使用两套严格度不同的数据集，对33种变异致病性预测程序及1种保守性评分进行了验证。结果显示，BayesDel（含addAF与noAF版本）、PON-P2（基因组版与蛋白质版）及ClinPred算法在两个子集的灵敏度、特异度、准确度及Kappa值上均表现最优。随后采用这5种算法对意义未明变异进行分析：结果表明，122个变异在5种预测工具中获得了一致的分类结果，其中57个被预测为致病性，65个被预测为良性；针对未被PON-P2覆盖的变异，其余4种预测工具一致将88个变异判定为致病性、92个判定为良性；剩余124个变异未获得预测工具的一致分类结果。