co-founder @drizzlehealth Sampling & concentrating bio samples across lab and industrial scales. prev @jhubme @iitbhu_varanasi
Baltimore/Bangalore
Joined December 2013
- Tweets876
- Following614
- Followers92
- Likes3.4K
Pinned Tweet
Most bioprocesses (food, fermenters, water treatment) are going to become model-controlled closed loops this decade. The same [Sense, decide, act, repeat] loop self-driving & robotics cracked.
But industrial bio is stuck a layer lower than people think, & the models aren't what's holding it back. (1/6)
Digvijay retweeted
Is there an implicit epistologic here - what returns the fastest rate of discovery/insight. With Physics, experimentalists became in a sense 'faster' than theorists after the 50s. I think with math, the idea that compute solves/will solve everything may also run into 'for what types of problems is conjecturing the fastest way of progress/insight".
To misattribute lord Kelvin, "there's nothing new in x..." may be close with AI but possibly be a cyclic function of history.
I don't know anyone in Orlando but we're here at the @IntFreshProduce expo and showing our first ever portable prototype for industrial scale microbiology. This one is for food processing plants!
Had a fun time hosting this weekend's microbiology meetup in Bengaluru! Fun people!
We discussed a lot of things around food production, amr, & even Navier-Stokes!
Some (meta) commentary I enjoyed
- how close we really are with AI engineered organisms for biosecurity
- the need for publication being an impediment for amr sequencing data being used for a real time Bloomberg terminal
- resistance levels for Ciprofloxacin approaching 90% in IPD
- using CFD in conversation is confusing when you have both biologists & engineers lol
- EV is everywhere. Farraz (who I didn't even know was at CCAMP) showed up!
Look forward to the next one!
This makes me think of @lukeburgis 's freshmanistan vs celebristan.
Unless there's social proximity, you won't compete with AI same way you can't compete with and want to become a doggo. Because in all estimations, AI will not compete with us for air, water, food, shelter, sleep & reproduction in the same short causal way that people intuitively understand.
It doesn't matter for most people that Elon or SpaceX exist - they're too far away in celebristan. Most people don't directly compete with Elon or SpaceX (or actively think of them for that matter).
One of my favorite songs is Sanson Ki Mala. It was composed by him but written by lesser known Tufail Hoshiarpuri (also from Pakistan). I found it cool that the song is about devotion but expressed through Mirabai's relationship with Krishna, which is essentially rooted in Hindu lore. i.e. Sufi & Bhakti movements had similar values, esp around surrender. Another song he did (Jogi de Naal) is similarly from Bulleh Shah (where Jogi is more wander-like).
I wish 8th grade history started with artifacts like this.
one more nusrat thing, bc I realize I’ve never shared one of the all-time bangers, the Yokohama recording of mustt mustt, which sounds as though it came from the ground, as much of the best music does
youtu.be/SDfELfpumEE?si=xBZC…
1810.11065v1.pdf share.google/TukhmCQZCxWePl3…
Those are not stars. Those are "hot pixels", damaged pixels in the camera sensor. What you are seeing is radiation damage and cosmic ray damage accumulated after the long trip.
Space is bad for your DNA. And it gets far worse near the big planets. Jupiter, fo example:
Ganymede: ~8 rem/day (Moderately high, shielded slightly by its own weak magnetic field
Europa: ~540 rem/day (Severe, sitting deep inside the harsh radiation belt)
Io: ~3,600 rem/day (Extreme, the highest radiation and volcanic activity)
Only Callisto (~0.01 rem/day, seated outside the main radiation belts) is viable for human exploration.
You know what I didn't model? Scratches.
Injection molding facility. fml
Did some math in the park again this week. I overestimated my abilities & barely moved one USAMO problem. Last year I'd been able to eventually do 2 RMO ones.
1. USAMO is also way more difficult than RMO per comments online.
2. The problems in both seem deceptively approachable.
3. JEE ones were pretty tedious & seemed to look for silly mistakes. Similar number as last year (had to look up formulas both times).
4. There are no Rajma Chawal places in Baltimore. Not even terrible ones. Why?
Actually the problem at the bottom is IMO.
Let n be an even positive integer (n = 2k). Two different cells of an n × n board are neighboring if they share a common side. Find the minimal number of cells on the n × n board that must be marked so that every cell (whether marked or unmarked) has at least one marked neighboring cell.
I did it last year for the first time. Fun way to also meet interesting people: xitter.cf/S_Digvj/status/1953863…
Not to toot our own horn, but this is exactly why @DrizzleHealth has built sensing on large physical volume samples! Our washwater testing system offers almost a million times more coverage than current testing!
Also, testing for pathogens in lettuce is extremely hard (& imo) random. You essentially grab a handful of sample from a room full of lettuce, say that it represents the entire thing and test that. In this case, that'd be catching lightning twice!
xitter.cf/S_Digvj/status/2028559…
Digvijay retweeted
100% agree the need for biosecurity for food! Our approach at @DrizzleHealth is large scale diagnostics. Genomics are inherently small volume (uL) making them cost & time (due to enrichment needs) prohibitive. Outbreaks like @McDonalds are proof. Screen EVERYTHING, PCR as needed
I don't know the details here obviously about why this pair brought in these pathogens -- but it does highlight that we don't just need biosecurity for humans -- we also need it for our food supply. COVID for corn would be more disruptive than a nuke hitting the US. We need to be biosecure and we are not.
The foundation for that is pathogen surveillance via environmental DNA sequencing in transport hubs and other obvious places (BSL4 labs, bio research institutes). Need BIORADAR collecting BIOINT so we get notice as early as possible to respond with countermeasures and also attribution so we can deter with threats. More in link in thread:
Digvijay retweeted
@YourTaylorFarms is actually not at fault - the industry uses outdated tech. tldr; most lettuce is not tested cus you're trying to minimize time at processing (to maximize shelf-life). An E coli test for full batch is unfeasible with current tech. It can take 2 days (enrichment), & ONLY tests <0.1% of the batch.
xitter.cf/S_Digvj/status/2028559…
Also why newer IITs shouldn't imitate the older ones, & the older ones shouldn't imitate the US Universities to climb the QS ranking ladder!
xitter.cf/S_Digvj/status/2065166…
