Hacker News Reader

Comment by new299 | Hacker News Reader

Right, I think we’ve probably missed boat on developing new diagnostic methods for Sars-CoV2.

In particular, this method appears to be antibody based? (Which has accuracy issues) and uses SPR, which may involve some technical risk.

However, I think there’s mileage in developing methods now for the next pandemic. My personal interest is in developing programmable qPCR-like systems [1]. So that kits can be deployed ahead of time, and then programmed to a specific target as required. If anyone is interested in discussing diagnostic approaches, please get in touch.

[1] http://41j.com/blog/2020/03/thoughts-on-a-new-approach-to-vi...

View on HN

I'm not familiar with the acronym PSM. Can you expand?

Are you familiar with the work of Dr. Chui at UCSF? His group has done some really cool work using mNGS to detect/diagnose emerging/rare infections in critically-ill patients with refractory encephalopathy

Sorry, typo. I meant SPR (surface plasmon resonance).

I’ve worked at a number of NGS platform companies developing new sequencing approaches. The problem is that sequencing is still expensive at the per-run level. It’s possible to be cost competitive with qPCR if you multiplex samples. But this isn’t ideal.

It would be interesting to create a small/cheap sequencer which could be applied to point-of-care/at-home testing. However, most of the money has gone after attacking the market leader (Illumina) on a cost-per-base, rather than cost-per-run.

A 1USD per-run sequencer would be interesting. But I’ve not seen anything that will hit that target in development. If anyone reading this is developing such a system, let me know, I’d love to get involved.

The idea of a programmable qPCR system is to add some of the versatility of sequencing to qPCR.

We've developed something NGS-based we call SwabSeq. You can find out more here: https://www.notion.so/Octant-SwabSeq-Testing-9eb80e793d7e463...

You can get to $1/sample; but need >1000 samples/run at least to get to that cost level. Could run 10k/day without automation; likely a lot more (100k-1MM) with automation.

Oh, nvmd. I just saw you wanted a $1/run. Something ONT based perhaps.

Reply on news.ycombinator.com