Genomic sequencing has become increasingly accessible as the cost of sequencing decreases. This has led to a rise in whole genome sequencing (WGS) and whole exome sequencing (WES), generating massive amounts of data. However, analyzing this data and extracting meaningful insights remains a challenge.
To address this challenge, researchers have turned to artificial intelligence (AI) methods, specifically genomic AI, which leverages large amounts of structured data paired with validated outcomes for training. Genomic AI has the potential to significantly reduce the time and effort required to analyze and interpret sequencing data, but it requires careful assembly of data across the entire analysis process.
One area where genomic AI is making strides is in variant calling accuracy. Illumina, a leading genomics company, has developed the DRAGEN™ secondary analysis pipeline, which uses machine learning algorithms to improve variant calling accuracy over a wide range of genomic regions. The latest release of DRAGEN, trained on vast amounts of data, achieves an analytical accuracy of 99.84%, reducing both false positive and false negative rates.
Another application of AI in genomic analysis is in predicting variant pathogenicity. Illumina scientists have developed PrimateAI-3D, a three-dimensional convolutional neural network that uses primate variants and 3-D protein structure to predict the effect of variants. This approach has been validated across six clinical benchmarks and has enabled the discovery of more significant gene-phenotype associations in rare variant burden tests.
In addition to protein-coding variants, Illumina has also developed SpliceAI, a deep learning model for identifying pathogenic variants in the non-coding genome. This extends clinical sequencing beyond the exome to the whole genome, improving the identification of disease-causing variants.
To aid in variant interpretation, Illumina has integrated Explainable AI (XAI) into their Emedgene™ tertiary analysis software. XAI prioritizes variants that are most likely to solve a case, providing valuable insights for clinical decision-making.
These advancements in genomic AI have the potential to revolutionize genomic analysis, enabling more accurate and efficient interpretation of sequencing data. As researchers continue to leverage AI and machine learning, we can expect further breakthroughs in understanding the complex nature of the human genome and its clinical implications.
The source of the article is from the blog crasel.tk