xyhopguy

Eh I disagree. It's been known for a while that you need something other than just DNA to get the heterogeneity of cellular phenotypes. Remember that your Iris and toenails have the same DNA, yet wildly different gene expression profiles. Without cellular memory that would be impossible and we would all still be single celled organisms :)

Maybe ten years ago I watched Bob Kennedy Jr or 3 or something talk about having battery swap stations instead of charging stations. Seemed interesting but I don't know much about batteries!

I think that's how you derive UCB, but optimizing cumulative regret rather than finding the probability distribution directly. Pls correct me if I'm wrong

Feels a lot more like NYC is an anomaly where no one cares about rules directly impacting them.

The Pacific Northwest in general seems to have a more communal culture (moreso Oregon than Western Washington). But NYC.... Dear lord.

Within 20 miles of somewhere like Portland for that matter. My in-laws live in the Portland burbs and everything between their house and the next town is farmland.

> (particularly the "go back" one).

Holy shit, once I built a web app for some biologists that wanted a "designer wizard." First I implemented a really shitty buggy wizard. Then I refactored it all into a state machine...........

I've also done this math -- price per sqft is wayy better in Seattle and you don't necessarily need a car. In any case, a metro card isnt cheap. The cost of groceries/alcohol is also much much higher in nyc and at a lower quality.

I have a postdoc friend who lives in Seattle on slightly less than I do. They have a huge 1br in cap Hill with parking next to the subway with a short committee downtown or to UW. Meanwhile my apartment is the same price with a minifridge, a 40m commute, and no dishwasher/laundry/disposal.

Plus the PNW is just more enjoyable to explore than the northeast in my opinion. You definitely can make more in Nyc but the COL and QoL is just awful compared to the pnw

Interestingly enough, I grew up in a small town outside Corvallis. Moving to a city as (relatively) big as Corvallis was pretty weird.

Bioinformatics is perpetually ten years behind. The de facto standard for sequencing data is effectively a stripped down bzipped plain text file. It's madness.

just freestyle, nice and relaxed. swimming distance isnt the best measure though, some people are more/less efficient than others.

used to do this during swim practice. unbelievably relaxing and amazing how far you can swim without a breath with a slow buildup. we used to go up to 4 lengths aka 100m

MCTS has always been popular and probably the best online/out-of-the-box MDP technique. But fundamentally its different than RL methods. In MCTS you assume you have a perfect simulation of the MDP which is rarely true outside of games. MCTS also doesnt really have a learning phase.

> My intuition is that MCTS is better for planning, but takes longer to train/evaluate.

There is no training phase in MCTS, rollouts can take a while, its important to have a fast simulator and rollout policy (like random/UCT).

> Is there some games one will work better than the other?

Games with simulators and perfect information!

I don't really know any history about cancer research, but as long as I've been in genomics people have talked about it. Difficulty is actually finding CNVs. These days it's not so hard.

It's also important to remember that we know some cancers (leukemia for instance) are diploid and rarely develop anueploidy.

Personally I see crispr more as a tool for functional genomics. It took like two decades to knockout all pairs of genes in yeast. With crispr I've seen people do it in a single experiment.

can't say too much but I work on the stats side of liquid biopsy. Been in genomics (on the cs side) for the last 6 years. the encouring part to me is 1) we can predict the effect of a treatment given the tumors clonal composition, 2) we can test the effects of drugs much more quickly thx to crispr 3) easily profile a tumors composition with cfDNA.

1 and 3 imply we may be able to drive clonal evolution to a place where we can indefinitely treat patients. 2 gives us a lot more options.

My (weak) understanding is that the functional impact of a mutation is more important than the type of variations. Otherwise we would see recurrent point mutations rather than hotspots right?

FWIW though, people have already been able to predict treatment outcome from mutation profile. Turns out the hard part is picking up on mutations present in only a tiny subset of the tumor.

Those already exist! TCGA for starters..

To be fair though, we are really close with cancer. Like really really close. cfDNA is revolutionary for our measurements. Crispr is revolutionary for understanding. So close.

It's worse than that. It turns imaging data (illumina basecaller) into sequences, aligns them, turns them back into images (nucleotides go into the same channel for some reason), THEN uses 100x more compute to get a questionable 1% gain on indels.

Why.

The biggest complaint about DeepVariant is that it isn't even a CV problem. I mean you start with a bunch of images, but those get processed and turned into basecalls, which get aligned to the genome, and then turned into an image again (why though?) for DeepVariant. Interesting approach to say the least.

even nature is hyping the verily paper now?

It's like I'm taking crazy pills. Variant calling was already very very good and grounded in statistics. Why ON EARTH do we need to convert it to AN IMAGE and give it 1000x more compute? The gains are like 1% too. That's barely above the error rate of sequencing.

The suburbs also have urban growth boundaries (parents live next to one) and it's easy to build over single family homes. The people and the city don't seem very interested however. The whole state is really antigrowth and it has been for a century.

there is also a mountain range in the way.

Dang sorry man that wasn't what I meant. The Ugb encourages density as you said, which is quite good if you want a comprehensive transit system.

maybe if they weren't so vulnerable to climate change. My money is on Portland. Good climate projects, tons of natural resources, strong urban growth boundry restricting sprawl. If the state ever figures it's shit out Portland would boom.

They will never be able to address the issues in New York. The whole place is polluted to hell -- the beaches are gross and dirty. I mean for fucks sake you have to ASK to not get a plastic bag (usually two). What year is it, 2002? The place literally smells like hot garbage and piss. The people have no respect for the land (ever go into prospect park on a summer weekend?). People working in business al

> You have almost literally the best of anything in the world within that hour-or-less subway ride.

Bullshit, You can take a 10 minute Bike/Light rail ride in portland and get to straight up farm-land or forest. In an hour you can get to the coast, desert, or mountains. In San Diego you can bike to the beach from anywhere in 10-20 minutes (that arent crowded and over trashed).

i wouldn't call new york an example of a "high quality of living." Subways that barely work? Terrifyingly chaotic traffic? A "me-first" attitude that is visible everywhere from the people complaining about citi-bikes to the advertisements literally EVERYWHERE.

Going from coast-to-coast is like walking into a time machine.