JMIR Publications

ABSTRACT

Background:

In recent years, thanks to the rapid development of next-generation sequencing (NGS) technology, an entire human genome can be sequenced in a short period of time. Therefore, NGS technology is being widely introduced into clinical diagnosis practice, especially with those diagnosis of hereditary disorders. Processing the DNA sequence data of a patient requires multiple tools and complex bioinformatics pipelines, and the exome data of single nucleotide variant (SNVs) will be generated.

Objective:

To assist physicians to interpret the genetic variation information generated by NGS in a short period of time

Methods:

We constructed a machine learning model for disease causing variants prediction in exome data. In our research, we collected sequencing data from whole exome sequencing and gene panel as training set. Then we integrated variant annotations from multiple genetic databases for model training. The model we built will rank SNVs and output the most possible disease-causing candidates. For model testing, we collected whole exome sequencing data from 108 patients with rare genetic disorders in National Taiwan University Hospital. We applied sequencing data and phenotypic information automatically extracted by keyword extraction tool from patient's electronic medical records into our machine learning model.

Results:

we succeed in 92.6% of the cases to locate the causative variant in the top 10 ranking list of average 741 candidate variants per person after filtering.

Conclusions:

The model ranks the same as manual performance, and it has been to use to help clinical diagnosis with genetic diseases.