Illumina has announced its plan to sell Grail, a cancer test developer, following a federal appeals court decision upholding the FTC's antitrust ruling against their $7.1 billion acquisition deal in 2021. This move comes after both U.S. and European Union regulators opposed the acquisition, citing concerns about diminished innovation and market competition. The sale of Grail, a company known for its early cancer detection technology, will occur through a 3rd party sale or capital market transaction.
Ion Stoica's lab continues to be a powerhouse of innovation. Previous successes of Stoica and his students include (but are certainly not limited to) Apache Spark, Ray, Apache Mesos and Alluxio.
The GRAIL press release [1] has more detailed stats.
> The analysis showed strong performance of GRAIL’s MCED methylation-based platform in the symptomatic population of more than 6,000 patients and demonstrated the feasibility of using an MCED test to assist clinicians with decisions around the route of referral from primary care.
> Within the study, 368 (6.7%) of the 5,461 evaluable patients were diagnosed with cancer through standard of care. The most common cancer diagnoses were colorectal (37.2%), lung (22.0%), uterine (8.2%), oesophago-gastric (6.0%) and ovarian (3.8%).
> GRAIL’s MCED test detected a cancer signal in 323 people, 244 in whom cancer was diagnosed, resulting in a positive predictive value (PPV) of 75.5%, negative predictive value (NPV) of 97.6%, and specificity of 98.4%. The overall sensitivity of the MCED test was 66.3%, ranging from 24.2% in stage I cancers to 95.3% in stage IV, and increased with age and later cancer stage. The overall accuracy of the top CSO prediction after a positive MCED test was 85.2%. The mean age of patients in the study was 62.1 years old.
Reposts are very common on Hacker News. I wonder if incorporating the performance of the same links with differing titles might've yielded a better training set.
Posting guidelines on HN generally require you to not editorialize titles, so most reposts have the exact same title (since it's the title of the original article).
I wanted a trimmed down interface for following live chess streams. To keep the website updated I run an AWS Lambda every two minutes that does the following:
1. Pulls active streams from the Twitch API.
2. Uses the Go templates library to repackage the response as static HTML.
3. Uploads the static HTML to S3, where it is served behind CloudFront.
Average latency for the Lambda over the last week is 4.84s and it runs with 128mb of memory. Running every two minutes, that comes to ~25,000 invocations and ~15,000 GB-seconds of compute time per month. The AWS Lambda free usage tier includes 1M free requests per month and 400,000 GB-seconds of compute time per month so the Lambda is definitely free.
The website is tiny (homepage < 20kb excluding the thumbnails which are hosted by Twitch), so I'm also inside the free tier on S3 and CloudFront. I paid $12 upfront for the domain registration.
