Editorial: Computational Learning Models and Methods Driven by Omics for Precision Medicine

Lei Zhu, Hongmin Cai, Fa Zhang, Quan Zou, Yanjie Wei, Huiru Zheng
2020 Frontiers in Genetics  
Due to the high experimental cost and the exponential decline in the cost of high-throughput sequencing, computational models, and methods are preferred by scholars. The curse of dimensionality is the primary obstacle to dealing with the explosive growth of omics data. Machine learning methods are applied to reduce dimensionality and perform feature selection from massive data. Researchers meet the requirements of data sparsity by increasing the sparsity constraints of the computational models.
more » ... The models combined with the deep learning method help to discover potential non-linear associations. Improving data representation or adding embedding layers could provide better performance of the models. Computational methods for biomarker discovery, sample classification, and disease process interpretation pave the way for precision medicine. This topic includes 34 papers and a corrigendum. These papers introduce latest researches in the area of computational biology, catering for precision medicine and complex diseases. They include sequencing alignment, correlation detection between omics data and biological traits, prediction of biological functionality, computational methods for cancer subtyping, finding of pathogenic causes, repositions and targeting, and computational methods specially designed for biological knowledge mining. SEQUENCE ALIGNMENT The raw sequencing data is unstructured short sequences. The structured data can be generated from downstream analysis through filtering, quality control, and assembly of these unstructured data. Assembly reconciliation can generate high-quality assembly results. In Tang et al., using the consensus blocks between contigs to construct adjacency graphs to avoid varying sequencing depth and sequencing errors, the authors propose a scoring function to rank the input assembly sets. They use an adjacency algebra model for accurate fusion, which performs well on M. abscessus, B. fragilis, R. sphaeroides, and V. cholerae. Shi and Zhang apply the partition and recur platform to generate a high-level abstraction of the sequence alignments. The algorithm component library is verified
doi:10.3389/fgene.2020.620976 pmid:33424938 pmcid:PMC7785880 fatcat:s4ya4eey7bh5ld7eqiae2hpoba