15 replies [Last post]
jeff101's picture
User offline. Last seen 3 hours 7 min ago. Offline
Joined: 04/20/2012
Groups: Go Science

Folks have been talking today in Vet Chat about using Neural Networks
to learn from Foldit players how to fold proteins.

GalaxyZoo is already doing things like this,
using players' results as training sets for Neural Networks,
as described in the article below:

[1] How Machine Vision Is Reinventing the Study of Galaxies
Thanks to Galaxy Zoo, crowdpower has revolutionized galaxy classification.
But now the machines are taking over.
http://www.technologyreview.com/view/536411/how-machine-vision-is-reinventing-the-study-of-galaxies/

Is the Foldit team trying to do anything like in the above paper?

jeff101's picture
User offline. Last seen 3 hours 7 min ago. Offline
Joined: 04/20/2012
Groups: Go Science
Another article:

The following article discusses an application of Neural Networks in protein folding:

[2] Reconstructing Protein Structures by Neural Network
Pairwise Interaction Fields and Iterative Decoy Set Construction
http://dx.doi.org/10.3390/biom4010160
http://www.mdpi.com/2218-273X/4/1/160

jeff101's picture
User offline. Last seen 3 hours 7 min ago. Offline
Joined: 04/20/2012
Groups: Go Science
Another article:

The following article discusses the efforts of "citizen scientists" in several Zooniverse projects.
It says "the Zooniverse team is working to use its results from crowdsourcing to train computers
to do more of the data analysis." It also discusses a new "build-it-yourself" tool
in which new projects "can be built in hours or days" instead of months.

[3] "Crowdsourcing platform gets results:
On the Zooniverse, data-analysis projects mesh public outreach and scientific discovery."
from pp.25-27 of the January 2016 issue of Physics Today
http://dx.doi.org/10.1063/PT.3.3047

jeff101's picture
User offline. Last seen 3 hours 7 min ago. Offline
Joined: 04/20/2012
Groups: Go Science
Maybe movies of the best-scoring solutions would help:

https://www.youtube.com/playlist?list=PLzgO42BPIbEDX1Jo78MYyo25P8nTIq6YS

has a collection of videos about old Foldit puzzles, like the one below:

CASP8 T0492 Start 3 Re-test (0:38 long)
https://www.youtube.com/watch?v=YN4tQ7tlDp0

These videos show how the protein changes as it is folded from its initial
structure to its final structure. Perhaps the inputs for a Neural Network
could be certain coordinates from the protein at time t, and the outputs
could be what sections were moved and how they were moved right after time t.

jeff101's picture
User offline. Last seen 3 hours 7 min ago. Offline
Joined: 04/20/2012
Groups: Go Science
Should the movies include false steps?

In the movie https://www.youtube.com/watch?v=YN4tQ7tlDp0
the score seems to always rise, and the protein undergoes smooth motions.
I suppose not all trajectories to reach the best fold will be this way.
I can imagine having to tolerate some loss of points to get out of a
local extremum (when the protein seems "stuck"). Also, many recipes
only manage to raise the score after a series of false steps. Finally,
some tools and hand-folding tricks make very jerky motions.

Would the best training trajectories be ones with the least number
of simple moves to go from start to finish? Would it be best if each
move raised the score? Would it be best if each move was a smooth one?

I'd imagine that real trajectories would need to be pruned a bit to
make good training trajectories, and some real trajectories probably
contain too many jerks or complicated moves to be useful as training
trajectories, even after some pruning.

Joined: 04/28/2015
Groups: Go Science
Thank you! Fold-it the

Thank you! Fold-it the project should really use and experiment with advanced technologies, including distributed neural network training with the best players, as well as tools for mutations of neural networks.

If the project folded knows all the best action of the players why those actions can not be programmed for the mutation of the best decisions?

This is a new feature in the client base and which has a self-study of the entire project fold it

jeff101's picture
User offline. Last seen 3 hours 7 min ago. Offline
Joined: 04/20/2012
Groups: Go Science
How to learn from heavy recipe users?

If certain players do almost no hand-folding,
maybe a Neural Network of Recipes could be made
from the series of Recipes these players choose.

First, go through all the Recipes and try to sort them into
a reasonable number of groups of closely-related Recipes. You
could use each Recipe's parent and children and each Recipe's
name to help sort them into groups. You could also compare the
actual text in their source code, sort of like looking for
sequence homology.

Next, record the sequence of Recipes used to go from the
starting structure to the best-folded structure. If a Recipe
did not improve the score, it could be omitted from the sequence.
The Neural Network's inputs could be the previous N Recipe groups
used, and the Network's output could be what Recipe group to use
next. The Network could learn which Recipe groups tend to come
first, which Recipe groups tend to follow other Recipe groups,
how many Recipes to use between each use of one from the DRW
group, how many Recipes to use between each use of one from
the GAB group, etc.

You might also include transitions between wiggle powers as
Network inputs. Then the Network could learn which Recipe groups
are best at each wiggle power, right after switching from low to
high wiggle power, etc.

jeff101's picture
User offline. Last seen 3 hours 7 min ago. Offline
Joined: 04/20/2012
Groups: Go Science
Do Neural Networks learn from both good and bad moves?

If a Neural Network can learn from actions that drop
the score as well as actions that improve the score,
it could learn a lot more from our results.
After all, most of what we (and our Recipes)
do fails to improve the score, especially
near the end of a puzzle.

To cite a line from the movie "I Origins"
(http://www.imdb.com/title/tt2884206/)
"Turning over rocks and finding nothing is progress".

jeff101's picture
User offline. Last seen 3 hours 7 min ago. Offline
Joined: 04/20/2012
Groups: Go Science
Could log.txt be used to make a Neural Network for each player?

Does log.txt contain enough information to make
a Neural Network for each player? Maybe some
clever coder could make a program to do this.
The program could be run every time the user
quits a Foldit client. I think log.txt gets
overwritten when a new Foldit client starts.

Joined: 04/28/2015
Groups: Go Science
The prototype of a neural

The prototype of a neural network for the best results in the self-study.
1) We do not have comprehensive statistics of players to the neural network to make better decisions.
2) The actions of all the players should be in the public domain for self-neural networks.
On a free basis, without charges and licenses

Joined: 04/28/2015
Groups: Go Science
concept

concept

jeff101's picture
User offline. Last seen 3 hours 7 min ago. Offline
Joined: 04/20/2012
Groups: Go Science
What do the above figures show?

What do the above figures show?
Did you make them yourself?
How did you make them?
If you used publicly available programs,
please provide links and references for them.

Joined: 04/28/2015
Groups: Go Science
This art design. Such

This art design.

Such programs do not exist at the moment

Joined: 04/28/2015
Groups: Go Science
hexagons used everywhere Six

hexagons used everywhere
Six points are obtained from the 3 segments
First of all, you need to look for a correlation with the number six, proportion, calculation results - the final structure

then maybe the results will be mathematically grounded

Joined: 04/28/2015
Groups: Go Science
If administrators allow me to

If administrators allow me to put the image of the universal mathematical law.
If all neurons are interconnected - all the numbers also interconnected
Then all the parts of the protein must also be linked. ( biochemistry )
all events take place across the different conditions of the transition. And if you create these conditions (different) you can get answers. You can consider it a theory, and it is possible truth.
no difference. It's all the same thing.

egran48's picture
User offline. Last seen 1 year 10 weeks ago. Offline
Joined: 03/31/2014
Groups: Go Science
Concept of Neural Networks

Thanks Jeff and 010 for drawing our attention to Neural Networks.

I can't comment on the maths but the concept of using Neural Networks sounds like it would be worthy of exploration. The idea of sorting recipes has merit, I have been trying to do this by trial and error for 18 months but would all players be willing to give up their advantages for the greater good?
If mathematics and the study of the natural world (algorithm for bee colony article)could speed up what Foldit is working towards then that has to be good.
I'll be interested to see where this goes....

Sitemap

Developed by: UW Center for Game Science, UW Institute for Protein Design, Northeastern University, Vanderbilt University Meiler Lab, UC Davis
Supported by: DARPA, NSF, NIH, HHMI, Microsoft, Adobe, RosettaCommons