On Maximum Likelihood Reconstruction over Multiple Deletion Channels
On Maximum Likelihood Reconstruction over Multiple Deletion Channels
复制标题
DOI：
10.1109/isit.2018.8437519
复制
发表时间：
2018-06
期刊：
影响因子：
--
通讯作者：
Sundara Rajan Srinivasavaradhan;M. Du;S. Diggavi;C. Fragouli
中科院分区：
文献类型：
--
作者：
Sundara Rajan Srinivasavaradhan;M. Du;S. Diggavi;C. Fragouli
在“从头开始”DNA测序中，当通过多个缺失通道观察时，重建序列的问题发生了。DNA可以被测序多次，产生它的几种“外观”，但每次测序器都可能受到（独立的）缺失损伤的干扰。本文的主要目标是为通过固定数量的删除通道透镜观察到的序列开发重建算法。我们使用删除通道的概率模型来开发符号和序列最大似然解码标准，以及由它们驱动的算法。数值评估表明，与早期算法相比，该算法在编辑距离误差方面有所改进，
The problem of reconstructing a sequence when observed through multiple looks over deletion channels occurs in “de novo” DNA sequencing. The DNA could be sequenced multiple times, yielding several “looks” of it, but each time the sequencer could be noisy with (independent) deletion impairments. The main goal of this paper is to develop reconstruction algorithms for a sequence observed through the lens of a fixed number of deletion channels. We use the probabilistic model of the deletion channels to develop both symbol-wise and sequence maximum likelihood decoding criteria, and algorithms motivated by them. Numerical evaluations demonstrate improvement in terms of edit distance error, over earlier algorithms,