4 ms·

AlQuraishi's tweet [0] about this: > CASP14 #s just came out and they’re astounding—DeepMind looks to have solved protein structure prediction. Median GDT_TS w

by xral 6y ago

AlQuraishi's tweet [0] about this: > CASP14 #s just came out and they’re astounding—DeepMind looks to have solved protein structure prediction. Median GDT_TS went from 68.5 (CASP13) to 92.4!!!! Cf. their 2nd best CASP13 struct scored 92.8 (out of 100). Median RMSD is 2.1Å. I think it's over https://predictioncenter.org/casp14/zscores_final.cgi https://predictioncenter.org/casp14/zscores_final.cgi [0]: https://twitter.com/MoAlQuraishi/status/1333383634649313280 https://twitter.com/MoAlQuraishi/status/1333383634649313280

matsemann 6y ago
What does that Å mean? Never seen our letter been used in a scientific context.
- flobosg 6y ago
  Ångström, a length unit. 1 Å = 0.1 nm.
- seslattery 6y ago
  It's the symbol for Angstrom, a unit of length 10^-10m https://en.wikipedia.org/wiki/Angstrom https://en.wikipedia.org/wiki/Angstrom
- smt1 6y ago
  It used a lot when systems are examined at the nano-scale. Metrification and creating a "fubini's theorem" for a specific problem to measure something (indeed category theory is useful for building a localized "global wire" with appropriate "gauges" of interesting where optimization methods will work (to achieve the non-equilibrium control-theoretic orient-folds of interesting of whatever "the soln" is) with enough "space" to "try" pull-backs and push-forwards as needed (for a class/family of physically analogous of data). I thinking looking at things trough Joseph Fourier's eyes is pretty englightening. He sees to have ideated both the heat transfer problem (and being able to apply modern methodology by forming distributed or sparse representations of it, then assessing the non-linear dynamics of it modern robotics and mathematics senses to it, which would be very much applying pfaffian dynamics to me, and being able to know about cohomogies is a blessing such that the appropriate physical effect where the maximum likelihood is constained). is important in both scale free systems, fibers of networks of systems that need to be localized (this is approximately global sections of global optimization but then model indentified), mass effect which require some sort of techno-economic analysis (think the climate resilience problem) and (historically, I think COVID will shift that) lack of progress towards applied coding in the life sciences vs information sciences. What's pretty surreal to me is that exploring (and documenting some of interesting blurs between fields), say like math, physics, statistics, computer sciences, signal processing, natural language (even of language of scientific discourse), renormalization methods, naturalizations, socializations, and what are global/local laws lets you almost do a approach it as a "reverse Robin Hood" problem.
- kolinko 6y ago
  0.1nm - approximately a size of an atom - used in organic chemistry often.
  - adw 6y ago
    Standard distance measure in most atomic-scale condensed-matter fields. Certainly inorganic crystallography/materials science/condensed matter physics.
elwell 6y ago
> https://predictioncenter.org/casp14/zscores_final.cgi https://predictioncenter.org/casp14/zscores_final.cgi `.cgi`... we've come full circle

by xral 6y ago

matsemann 6y ago

What does that Å mean? Never seen our letter been used in a scientific context.

flobosg 6y ago
Ångström, a length unit. 1 Å = 0.1 nm.
seslattery 6y ago
It's the symbol for Angstrom, a unit of length 10^-10m https://en.wikipedia.org/wiki/Angstrom https://en.wikipedia.org/wiki/Angstrom
smt1 6y ago
It used a lot when systems are examined at the nano-scale. Metrification and creating a "fubini's theorem" for a specific problem to measure something (indeed category theory is useful for building a localized "global wire" with appropriate "gauges" of interesting where optimization methods will work (to achieve the non-equilibrium control-theoretic orient-folds of interesting of whatever "the soln" is) with enough "space" to "try" pull-backs and push-forwards as needed (for a class/family of physically analogous of data). I thinking looking at things trough Joseph Fourier's eyes is pretty englightening. He sees to have ideated both the heat transfer problem (and being able to apply modern methodology by forming distributed or sparse representations of it, then assessing the non-linear dynamics of it modern robotics and mathematics senses to it, which would be very much applying pfaffian dynamics to me, and being able to know about cohomogies is a blessing such that the appropriate physical effect where the maximum likelihood is constained). is important in both scale free systems, fibers of networks of systems that need to be localized (this is approximately global sections of global optimization but then model indentified), mass effect which require some sort of techno-economic analysis (think the climate resilience problem) and (historically, I think COVID will shift that) lack of progress towards applied coding in the life sciences vs information sciences. What's pretty surreal to me is that exploring (and documenting some of interesting blurs between fields), say like math, physics, statistics, computer sciences, signal processing, natural language (even of language of scientific discourse), renormalization methods, naturalizations, socializations, and what are global/local laws lets you almost do a approach it as a "reverse Robin Hood" problem.
kolinko 6y ago
0.1nm - approximately a size of an atom - used in organic chemistry often.
- adw 6y ago
  Standard distance measure in most atomic-scale condensed-matter fields. Certainly inorganic crystallography/materials science/condensed matter physics.

elwell 6y ago

> https://predictioncenter.org/casp14/zscores_final.cgi https://predictioncenter.org/casp14/zscores_final.cgi `.cgi`... we've come full circle