In cancer, recurrently mutated sites in DNA and proteins, called hotspots, are thought to be raised by positive selection and therefore important due to its potential functional impact. Although recent evidence for APOBEC enzymatic activity have shown that specific types of sequences are likely to be false, the identification of putative hotspots is important to confirm either its functional role or its mechanistic bias. In this work, an algorithm and a statistical model is presented to detect hotspots. The model consists of a beta-binomial component plus fixed effects that efficiently fits the distribution of mutated sites. The algorithm employs an optimal stepwise approach to find the model parameters. Simulations show that the proposed algorithmic model is highly accurate for common hotspots. The approach has been applied to TCGA mutational data from 33 cancer types. The results show that well-known cancer hotspots are easily detected. Besides, novel hotspots are also detected. An analysis of the sequence context of detected hotspots show a preference for TCG sites that may be related to APOBEC or other unknown mechanistic biases. The detected hotspots are available online in http://bioinformatica.mty.itesm.mx/HotSpotsAnnotations.

在癌症中，DNA 和蛋白质中反复突变的位点（称为热点）被认为是通过正选择产生的，因此由于其潜在的功能影响而很重要。尽管 APOBEC 酶活性的最新证据表明特定类型的序列可能是错误的，但假定热点的识别对于确认其功能作用或其机械偏差非常重要。在这项工作中，提出了一种算法和统计模型来检测热点。该模型由β 二项式组件和有效拟合突变位点分布的固定效应组成。该算法采用最优逐步方法来寻找模型参数。仿真表明，所提出的算法模型对于常见热点具有很高的准确性。该方法已应用于 33 种癌症类型的 TCGA 突变数据。结果表明，众所周知的癌症热点很容易被发现。此外，还检测到新的热点。对检测到的热点的序列上下文的分析表明，对 TCG 位点的偏好可能与 APOBEC 或其他未知的机械偏差有关。检测到的热点可在http://bioinformatica.mty.itesm.mx/HotSpotsAnnotations中在线获取。