The Lu Hui research group of Shanghai Jiaotong University has developed a new method for quality control of genomic next-generation sequencing data

Recently, the internationally renowned journal "Communications Biology" published the latest research results
of Professor Lu Hui's research group of Shanghai Jiaotong University entitled "FVC as an adaptive and accurate method for filtering variants from popular NGS analysis pipelines" 。 This study proposes a new quality control method for genomic next-generation sequencing data, which effectively improves the reliability
of genetic variants detected by next-generation sequencing.
Ren Yongyong and Kong Yan, doctoral graduates of the School of Life Science and Technology of Shanghai Jiao Tong University, are the co-first authors, and Professor Lu Hui and Professor Zhao Hongyu of the School of Life Science and Technology are the co-corresponding authors
.

Whole genome second-generation sequencing technology has been widely used in genomic research fields such as pediatric genetic disease diagnosis, tumor genome map analysis, and whole genome association analysis, but due to sequencing and analysis errors, there are a large number of false positive variants in the detected SNV and INDEL variant types, which brings great noise interference
to genome map analysis and differential diagnosis of genetic diseases 。 At present, there are multiple methods that can be used to filter the false-positive variants detected by next-generation sequencing, such as Frequency, Hard-Filter, VQSR, GARFIELD, and VEF, but these methods lose several times the true variant while filtering the false-positive variant, seriously interfering with the downstream diagnosis of genetic diseases and molecular function research
.

To solve the above problems, Lu Hui's team built an adaptive filtering method FVC (Filtering for Variant Calls) to filter
false positive variants in the detection results of different genetic variant analysis software (such as GATK HaplotypeCaller, Mutect, Varscan, and DeepVariant).

The results of the study showed that compared with other filtering methods, the new method FVC filtered out more false positive variants, and recalled ~51-99% of true positive variants missed by other filtering methods, when using the evaluation index OFO (Odds of false omission, the ratio of the number of true positive variants lost to the number of false positive variants filtered, also known as true positive loss ratio) for performance evaluation.
FVC reduced the true positive loss ratio OFO from 0.
05-1661.
28 to 0.
02-0.
57
.
To rule out potential data leakage and evaluation bias caused by overfitting, FVC also achieved the best performance
using leave-one-chromosome-out cross-validation, leave-one-individual-out cross-validation, and independent test sets, respectively.

This work was supported
by the High Performance Computing Center (HPC) of Shanghai Jiao Tong University, the Medical Research Fund of Shanghai Jiao Tong University, and the National Key Research and Development Program of China (2018YFC0910500).

Links to papers: