🚨 Google just shocked the world.

In partnership with

The Gold standard for AI news

AI keeps coming up at work, but you still don't get it?

That's exactly why 1M+ professionals working at Google, Meta, and OpenAI read Superhuman AI daily.

Here's what you get:

• Daily AI news that matters for your career - Filtered from 1000s of sources so you know what affects your industry.

• Step-by-step tutorials you can use immediately - Real prompts and workflows that solve actual business problems.

• New AI tools tested and reviewed - We try everything to deliver tools that drive real results.

• All in just 3 minutes a day

Join 1M+ pros

That means earlier detection, faster treatment, and survival rates we’ve never seen before.

Here’s how it works:

Cancer starts in your DNA.

Every cancer is basically a genetic bug.

One cell’s DNA mutates in a way that breaks the rules of cell division and suddenly it starts multiplying out of control.

Finding those mutations is the key to stopping cancer.

But it’s harder than it sounds.

When scientists sequence a tumor’s genome, they get billions of data points most of them noise.

Real mutations are buried under errors from the sequencing process itself.

Spotting which ones actually cause cancer is like finding one typo in a billion-character book.

DeepSomatic fixes that.

Developed by Google Research and UC Santa Cruz, DeepSomatic uses a convolutional neural network (CNN) the same kind of AI used for image recognition.

Except here, it’s not looking at cats or faces…

It’s looking at the human genome.

How it works:

Instead of reading DNA as text, DeepSomatic converts sequencing data into images.

Each pixel represents a piece of the genetic code and its alignment quality.

Then, the AI “sees” patterns that reveal whether a mutation is real — or just a sequencing glitch.

Trained on world-class data...

Google built a new dataset called CASTLE (Cancer Standards Long-read Evaluation).

It includes tumor and normal samples from:

4 breast cancers

2 lung cancers

Sequenced using 3 platforms: Illumina, PacBio, and Oxford Nanopore

This cross-platform design made DeepSomatic ultra-robust.

It even works on damaged samples.

Old tumor samples preserved in wax (FFPE) are notoriously messy.

DeepSomatic still nailed it finding real mutations other tools missed.

Even on exome-only data (just 1% of the genome), accuracy held strong.

That’s massive for real clinical use.

Real-world proof.

They tested it on:

- Glioblastoma (aggressive brain cancer)

- Pediatric leukemia (the most common childhood cancer)

Result: it detected all known mutations plus new ones that hadn’t been catalogued.

That’s discovery, not just diagnosis.

Google open-sourced this.

Both the DeepSomatic model and the CASTLE dataset are now available to researchers everywhere.

That means labs around the world can accelerate cancer variant discovery instantly.

Open science at its best.