How data science can make your genome actionable
summary
Company Freenome shows data pipelines that transform whole‑genome reads into image‑like representations and train neural nets to classify cancer vs non‑cancer with features drawn from TCGA‑style nucleotide strings, and to classify early vs late stage disease. They report preliminary classifiers across ovarian, breast, lung, prostate and colorectal cancer, a case study where a patient’s molecular age jumps with a 2011 mutation and genomic instability peaks on circos plots as disease progresses. Google’s variant‑calling work is cited as a parallel success.
AI-written summary · about
chapters (3)
related talks
57mBig Data and New Models Needed to Study DNA Variation in Evolution and Cancer
50mCancer Genomics - a biologist and a developer
01h 31mIntroduction to Cancer Bioinformatics I: Inferring Genomic Variation from Tumor Sequencing Data
35mFrom cells to drug responses - machine learning in cancer research
18mCan AI make us smarter than cancer
32mHow machine learning helps cancer research
38mInterpreting Deep Learning Models Of Functional Genomics Data To Decode Regulatory Sequence...
17mBig data in personalized medicine