Today's commercial liquid biopsies, like most tissue biopsies, rely on identifying known specific biomarkers, or detecting known or newly discovered somatic mutations in cancer high-frequency mutations, or in some cases. Detection of mutations in the entire exome. But Freenome said that these detection strategies have abandoned most of the genomic data that helped us identify cancer at the earliest.
Although he declined to disclose details of the company's genome-wide sequencing and training algorithms, Otte said he and his colleagues believe that their unique approach enables the analysis of whole-genome sequencing data at a practical price. Although they have not shared the data publicly, it seems that this method is effective.
Freenome's AGE-based liquid biopsy strategy
Their strategy is this: genome-wide sequencing of healthy human samples and cancer patient samples, submitting data to AGE until the system learns how to distinguish them. The more samples the AGE analyzes, the stronger the ability to distinguish between the cancer and non-cancer groups. Once the system is strong enough for clinical testing, Freenome will be able to sequence and classify customers' circulating free DNA (cfDNA).
Otte said, "In fact, this is something that machines can detect and track and humans can't do. For example, if you do mutation screening, the usual method is to match the reads to a certain location, and the same site will appear. The number of reads is 'A', and a certain number of reads is 'G'. If the site on the reference genome is A, and we detect a large amount of read G, which exceeds the set intercept point, then we think this The site contains a mutation."
"But because this intercept point can be set arbitrarily, if the value does not reach the threshold, then the entire data set will be discarded, even if only one read is lower than required. The meaning of this data is human understandable. But the machine can track more In many cases, whether a read is 'G' or 1000 reads are 'G', the machine can record the trace."
If a variation in the data is a sequencing error, the learning engine will filter the sufficient sample after it has been analyzed because the errors are random.
Surgical Tape,Adhesive Flexible Tape,Waterproof Surgical Tape,Breathable Surgicaltape
Zhende Medical Co.,Ltd , https://www.zhendemedicals.com