PyEMMA
PyEMMA copied to clipboard
markovmodel
Implement Laio density based clustering algorithm
This would be nice to have for estimating core MSMs. One can find an impl for R here: https://github.com/thomasp85/densityClust
@giopina Frank told me, that you are working on this or might be interested.
https://github.com/cwehmeyer/pydpc This algorithm needs user interaction, but also be interesting (for smaller data sets).
I'm working on this. Now they have an unsupervised version that works without the user selection of cluster centers that was required in the original algorithm. I have an implementation that I'm using. It's "PyEmma friendly" but independent, at the moment. For sure I'm interested into put it into PyEmma: what if we speak about it in one month? (I'll be back in FU from January) ps: if you need to use it now for some application I can share my code with you
Sure, this will be more efficient once you are here. I suggest that we build a working core MSM example with sklearn's dbscan clustering for now. Once this is in place, it should be easy to exchange the clustering algorithm.
@marscher, how about building a notebook where we do TICA + dklearn-dbscan + core MSM estimation using a toy model (double well) and a DESRES example (e.g. BPTI or one of Simon's favorites). That should be easy to do, we can test things that way, and write a little publication about the impl. Giovanni can help with the application in case if I'm unresponsive.
Am 01/12/16 um 16:54 schrieb giopina:
I'm working on this. Now they have an unsupervised version that works without the user selection of cluster centers that was required in the original algorithm. I have an implementation that I'm using. It's "PyEmma friendly" but independent, at the moment. For sure I'm interested into put it into PyEmma: what if we speak about it in one month? (I'll be back in FU from January) ps: if you need to use it now for some application I can share my code with you
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/markovmodel/PyEMMA/issues/998#issuecomment-264210874, or mute the thread https://github.com/notifications/unsubscribe-auth/AGMeQp78lhhcNbk6Ka1-UfZQkLngeal2ks5rDu2_gaJpZM4LAY3s.
--
Prof. Dr. Frank Noe Head of Computational Molecular Biology group Freie Universitaet Berlin
Phone: (+49) (0)30 838 75354 Web: research.franknoe.de Mail: Arnimallee 6, 14195 Berlin, Germany
Thanks for the feedback. I'm currently working on the wrapper for sklearn. @giopina we can work out the details, when you're here. That would be easier.
Very interesting! I read Laio's paper a while ago and ever since I would like to see an implementation of it in python to test. Can you tell me what's your status on it? Do the corresponding objects follow the scikit-learn clustering library?
