Menu
▾
▴
neuroml-python — All things Python related, particularly the Python API for NeuroML
You can subscribe to this list here.
| 2012 |
Jan
|
Feb
|
Mar
|
Apr
(1) |
May
(21) |
Jun
(11) |
Jul
(1) |
Aug
|
Sep
(1) |
Oct
|
Nov
|
Dec
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2013 |
Jan
|
Feb
(2) |
Mar
(1) |
Apr
(2) |
May
|
Jun
|
Jul
|
Aug
|
Sep
(1) |
Oct
|
Nov
(9) |
Dec
|
| 2016 |
Jan
|
Feb
(1) |
Mar
|
Apr
|
May
|
Jun
|
Jul
|
Aug
|
Sep
|
Oct
|
Nov
|
Dec
|
| 2017 |
Jan
|
Feb
|
Mar
|
Apr
|
May
|
Jun
|
Jul
|
Aug
|
Sep
|
Oct
(3) |
Nov
|
Dec
|
|
From: Padraig G. <p.g...@uc...> - 2012-06-22 13:55:08
|
Hi, I've put 9am to 6pm to the agenda, but some people are arriving later (Andrew is just lunchtime Mon to lunchtime Tues). We have the room till 6 but will see how long we last... I'll be there from about 8:30 and I'll tell the security guy on the door to let you sign in & just walk up to the 3rd floor. Any problems, call me on +44 797 9970283. Sandra Berger, who's in Sharon's group, but currently in Germany will be joining us too. Lunch & coffee are booked for the room. Dinner on Monday will be 19:30 BamBou: http://www.bam-bou.co.uk. Try to get Eduroam set up on your laptops or you'll be doing the meeting without an internet connection... The Skype call/Google Hangout will be held on Tues at 9:30 to get an update from Bangalore about MOOSE. Anything else? Re the code, I've added a dir https://github.com/NeuralEnsemble/libNeuroML/tree/master/hhExample which has a modified form of Andrew's suggested API, along with (placeholders for) the equivalent Python scripts in NEURON, MOOSE, Morphforge, and the current NML2 API generated from the Schema. These should all merge to be identical in an ideal world... Padraig P.S. Mike V & Michele, the cc'd copy of your messages to neu...@li... didn't get through as your gmail accounts aren't subscribed to the list. I've dumped those messages as everyone relevant for the meeting is on this list. On 22/06/12 13:42, Mike Vella wrote: > I suggest 9.00AM, unless some haven't reached London yet by then? > > All the best, > Mike > > On 22 June 2012 13:40, Michele Mattioni<mat...@gm...> wrote: >> Hello there, >> >> The time of the meeting is not there. What time does it start? >> >> Cheers, >> Michele. >> >> On Mon, Jun 18, 2012 at 5:38 PM, Padraig Gleeson<p.g...@uc...> wrote: >>> Dear all, >>> >>> There is a provisional agenda for the mini workshop in London on the >>> libNeuroML Python API online here: >>> http://libneuroml.readthedocs.org/en/latest/meeting_june_2012.html. >>> >>> Meeting attendees: please update/comment on the agenda by editing the >>> source: >>> https://github.com/NeuralEnsemble/libNeuroML/blob/master/doc/meeting_june_2012.txt >>> >>> All other interested parties: This agenda will be updated to become the >>> minutes during the meeting with agreements/status updates/new code& >>> examples. >>> >>> Regards, >>> Padraig >>> >>> ----------------------------------------------------- >>> Padraig Gleeson >>> Room 321, Anatomy Building >>> Department of Neuroscience, Physiology& Pharmacology >>> University College London >>> Gower Street >>> London WC1E 6BT >>> United Kingdom >>> >>> +44 207 679 3214 >>> p.g...@uc... >>> ----------------------------------------------------- >>> >>> >>> >>> ------------------------------------------------------------------------------ >>> Live Security Virtual Conference >>> Exclusive live event will cover all the ways today's security and >>> threat landscape has changed and how IT managers can respond. Discussions >>> will include endpoint security, mobile security and the latest in malware >>> threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ >>> _______________________________________________ >>> Neuroml-python mailing list >>> Neu...@li... >>> https://lists.sourceforge.net/lists/listinfo/neuroml-python >> ------------------------------------------------------------------------------ >> Live Security Virtual Conference >> Exclusive live event will cover all the ways today's security and >> threat landscape has changed and how IT managers can respond. Discussions >> will include endpoint security, mobile security and the latest in malware >> threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ >> _______________________________________________ >> Neuroml-python mailing list >> Neu...@li... >> https://lists.sourceforge.net/lists/listinfo/neuroml-python -- ----------------------------------------------------- Padraig Gleeson Room 321, Anatomy Building Department of Neuroscience, Physiology& Pharmacology University College London Gower Street London WC1E 6BT United Kingdom +44 207 679 3214 p.g...@uc... ----------------------------------------------------- |
|
From: Mike V. <vel...@gm...> - 2012-06-22 12:42:30
|
I suggest 9.00AM, unless some haven't reached London yet by then? All the best, Mike On 22 June 2012 13:40, Michele Mattioni <mat...@gm...> wrote: > Hello there, > > The time of the meeting is not there. What time does it start? > > Cheers, > Michele. > > On Mon, Jun 18, 2012 at 5:38 PM, Padraig Gleeson <p.g...@uc...> wrote: >> Dear all, >> >> There is a provisional agenda for the mini workshop in London on the >> libNeuroML Python API online here: >> http://libneuroml.readthedocs.org/en/latest/meeting_june_2012.html. >> >> Meeting attendees: please update/comment on the agenda by editing the >> source: >> https://github.com/NeuralEnsemble/libNeuroML/blob/master/doc/meeting_june_2012.txt >> >> All other interested parties: This agenda will be updated to become the >> minutes during the meeting with agreements/status updates/new code & >> examples. >> >> Regards, >> Padraig >> >> ----------------------------------------------------- >> Padraig Gleeson >> Room 321, Anatomy Building >> Department of Neuroscience, Physiology& Pharmacology >> University College London >> Gower Street >> London WC1E 6BT >> United Kingdom >> >> +44 207 679 3214 >> p.g...@uc... >> ----------------------------------------------------- >> >> >> >> ------------------------------------------------------------------------------ >> Live Security Virtual Conference >> Exclusive live event will cover all the ways today's security and >> threat landscape has changed and how IT managers can respond. Discussions >> will include endpoint security, mobile security and the latest in malware >> threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ >> _______________________________________________ >> Neuroml-python mailing list >> Neu...@li... >> https://lists.sourceforge.net/lists/listinfo/neuroml-python > > ------------------------------------------------------------------------------ > Live Security Virtual Conference > Exclusive live event will cover all the ways today's security and > threat landscape has changed and how IT managers can respond. Discussions > will include endpoint security, mobile security and the latest in malware > threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ > _______________________________________________ > Neuroml-python mailing list > Neu...@li... > https://lists.sourceforge.net/lists/listinfo/neuroml-python |
|
From: Michele M. <mat...@gm...> - 2012-06-22 12:40:50
|
Hello there, The time of the meeting is not there. What time does it start? Cheers, Michele. On Mon, Jun 18, 2012 at 5:38 PM, Padraig Gleeson <p.g...@uc...> wrote: > Dear all, > > There is a provisional agenda for the mini workshop in London on the > libNeuroML Python API online here: > http://libneuroml.readthedocs.org/en/latest/meeting_june_2012.html. > > Meeting attendees: please update/comment on the agenda by editing the > source: > https://github.com/NeuralEnsemble/libNeuroML/blob/master/doc/meeting_june_2012.txt > > All other interested parties: This agenda will be updated to become the > minutes during the meeting with agreements/status updates/new code & > examples. > > Regards, > Padraig > > ----------------------------------------------------- > Padraig Gleeson > Room 321, Anatomy Building > Department of Neuroscience, Physiology& Pharmacology > University College London > Gower Street > London WC1E 6BT > United Kingdom > > +44 207 679 3214 > p.g...@uc... > ----------------------------------------------------- > > > > ------------------------------------------------------------------------------ > Live Security Virtual Conference > Exclusive live event will cover all the ways today's security and > threat landscape has changed and how IT managers can respond. Discussions > will include endpoint security, mobile security and the latest in malware > threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ > _______________________________________________ > Neuroml-python mailing list > Neu...@li... > https://lists.sourceforge.net/lists/listinfo/neuroml-python |
|
From: Padraig G. <p.g...@uc...> - 2012-06-18 16:39:26
|
Dear all, There is a provisional agenda for the mini workshop in London on the libNeuroML Python API online here: http://libneuroml.readthedocs.org/en/latest/meeting_june_2012.html. Meeting attendees: please update/comment on the agenda by editing the source: https://github.com/NeuralEnsemble/libNeuroML/blob/master/doc/meeting_june_2012.txt All other interested parties: This agenda will be updated to become the minutes during the meeting with agreements/status updates/new code & examples. Regards, Padraig ----------------------------------------------------- Padraig Gleeson Room 321, Anatomy Building Department of Neuroscience, Physiology& Pharmacology University College London Gower Street London WC1E 6BT United Kingdom +44 207 679 3214 p.g...@uc... ----------------------------------------------------- |
|
From: Padraig G. <p.g...@uc...> - 2012-06-12 16:46:35
|
Hi all, I've done some consolidation of the project docs into the docs folder on GitHub, which gets automatically generated here: http://libneuroml.readthedocs.org/en/latest/index.html I've moved the "Scope of the project" file here as an intro to the goals of the API, so please have a read over it to see if you still agree with what's there: http://libneuroml.readthedocs.org/en/latest/scope_of_project.html I've also moved the "Tools relevant for this initiative" from the Wiki here too: http://libneuroml.readthedocs.org/en/latest/tools.html An additional document which I feel we need to agree on (particularly Mike V) is one summarising the concepts of Node, Segment and Section: http://libneuroml.readthedocs.org/en/latest/nodes_segments_sections.html. This tries to outline the usage of these terms by various tools/formats and come to a consensus on how they should be used in NeuroML & the API. It also outlines some (of the no doubt numerous) issues which arise from the proposed form in the NeuroML object model. It should provide a basis for the debate about whether to have the language/API enforce best practices in morphology specification, or support all scenarios and force parsing applications to decide what to do with the edge cases. Regards, Padraig -- ----------------------------------------------------- Padraig Gleeson Room 321, Anatomy Building Department of Neuroscience, Physiology& Pharmacology University College London Gower Street London WC1E 6BT United Kingdom +44 207 679 3214 p.g...@uc... ----------------------------------------------------- |
|
From: Andrew D. <and...@un...> - 2012-06-08 10:52:05
|
Hi everyone, I think it would be helpful for potential users of Python libNeuroML (and/or the multi-compartmental neuron simulation API (working name "pyramidal") that Mike Vella is working on for his GSoC project) to give examples of how they might want to use these packages. Here is are some examples of how I see multi-compartmental neurons in PyNN (in case the indentation is screwed up, also see https://gist.github.com/2762961). Note that these are just ideas: nothing is fixed in stone in my thinking on this: # Example 1 import pyNN.neuron as sim from neuroml import load_morphology, load_lems, Section from pyramidal.spatial_distributions import uniform, by_distance from nineml.abstraction_layer.readers import XMLReader from quantities import S, cm, um from pyNN.space import Grid2D, RandomStructure, Sphere pkj_morph = load_morphology("http://neuromorpho.org/...") # smart inference of morphology file format na_channel = XMLReader.read("na.xml") kdr_channel = load_lems("kd.xml") ka_channel = XMLReader.read("ka.xml") ampa = XMLReader.read("ampa.xml") gabaa = sim.ExpSynCond # first we define cell types (templates) purkinje_cell = sim.MultiCompartmentNeuron( morphology=pkj_morph, ion_channels={'na': na_channel, 'kdr': kdr_channel, 'kA': kA_channel}, channel_distribution={'na': uniform('all', 0.1*S/cm**2), 'kdr': by_distance(lambda d: 0.05*S/cm**2*d/200.0*um), 'kA': uniform('soma', 0.02*S/cm**2)} # need to specify cm, Ra (possibly with spatial variation) somewhere here synapses={'AMPA': ampa, 'GABA_A': gabaa}, synapse_distribution={'AMPA': uniform('all', 0.5/um), 'GABA_A': by_distance(lambda d: d<50.0)} ) gc_soma = Section(length=50, diam=10, label="soma") gc_dend = Section(length=200, diam=2, label="dendrite") gc_dend.connect(gc_soma) granule_cell = sim.MultiCompartmentNeuron(morphology=soma.get_morphology(), cm=…., Ra=...) granule_cell.soma.insert(na_channel, 0.05*S/cm**2) # using an imperative style of positioning ion channels granule_cell.soma.insert(kdr_channel, 0.01*S/cm**2) granule_cell.dendrite.insert(gabaa, label="GABA_A") # now we actually create the cells in the simulator purkinje_cells = sim.Population(100, purkinje_cell, initial_values={'v': -60.0}, structure=Grid2D(…)) granule_cells = sim.Population(1000, granule_cell, structure=RandomStructure(boundary=Sphere(radius=300.0))) # define which variables to record (purkinje_cells + granule_cells).record('spikes') granule_cells.sample(20).record('v') granule_cells[0:5].record('GABA_A.i', sections=['dendrite']) purkinje_cells[0].record('na.m', sections=longest_dendrite(pkj_morph)) # connect populations depressing = sim.SynapseDynamics(fast=sim.TsodysMarkramMechanism(U=500.0)) p2g = sim.Projection(purkinje_cells, granule_cells, sim.FixedProbabilityConnector(weights="0.1*exp(-d/100.0)", delays="0.2+d/100.0"), source="soma.v", target="GABA_A", synapse_dynamics=depressing) g2p = sim.Projection(granule_cells, purkinje_cells, sim.FromFileConnector("connections.h5"), source="soma.v", target="AMPA") sim.run(10000) (purkinje_cells + granule_cells).write_data("output.h5") sim.end() # ------------------------------- # Example 2 from pyNN.neuroml import read import pyNN.moose as sim sim.setup() network = read("complete_neuroml_model.xml") network.populations["purkinje"].record('spikes') network.populations["purkinje"].sample(10).record('v') sim.run(1000.0) data = network.get_data() sim.end() # ------------------------------- I'd be interested in seeing some other examples using your own preferred syntax and use-cases. Cheers, Andrew |
|
From: Padraig G. <p.g...@uc...> - 2012-05-29 17:56:08
|
Hi all, Andrew Davison, Mike Vella and I are planning a small face to face meeting to try to move the libNeuroML (https://github.com/NeuralEnsemble/libNeuroML) development forward. This is planned to take place at UCL on the 25th & 26th (Mon & Tues) June. Anyone interested in contributing to this (or just want to show off their Python based application in this general area) is very welcome to join. Please mail us for more details/logistics. There may be some support available for travel to this, but this is not yet confirmed. Please let us know if this is something you might require for attendance. Regards, Padraig ----------------------------------------------------- Padraig Gleeson Room 321, Anatomy Building Department of Neuroscience, Physiology& Pharmacology University College London Gower Street London WC1E 6BT United Kingdom +44 207 679 3214 p.g...@uc... ----------------------------------------------------- |
|
From: Michele M. <mat...@gm...> - 2012-05-28 15:26:39
|
Hello people, just to make sure everyone is aware (it's written in the README, but can be easily overlooked) the docs for the latest python effort for libneuroml is created using sphinx (classic python doc tools) and sits here: http://readthedocs.org/docs/libneuroml/ As soon new documentation is pushed in the doc folder https://github.com/NeuralEnsemble/libNeuroML/tree/master/doc read-the-docs rebuilds the website. Everybody can contribute to this doc, so should be easy to add things. All you need to do is to write the doc (ehm..), and push it to the master. Mike you could just create a new page called Reports and write there the summary of the biweekly meeting there. You just need to write ReST doc, you can have a quick view to the how_to_controbute_page https://github.com/NeuralEnsemble/libNeuroML/blob/master/doc/how_to_contribute.txt This could be one of the option. Cheers, Michele. |
|
From: Padraig G. <p.g...@uc...> - 2012-05-25 09:21:13
|
Hi, I agree than many applications will be interested in only the morph part of this API, and that is an argument for making a separate smaller package just for morphologies. However, these applications should also be able to read NeuroML files/HDF5 files containing additional info on channels/synapses/network connections. I think having all of this functionality included in one tightly integrated package will make it easier to exchange "full" neuron models between applications with a minimum loss of information. It might also encourage visualisation applications to allow display of channel densities etc. (in the same way as groups) even if they can't simulate the cell. Re use of the package as a simulator front end or not: the package should definitely be able to be used standalone for loading in/parsing morphs/networks/etc. The hooks to allow a simulator to be used as the "backend" for data storage could be present to a certain extent in this package, but probably PyNN is the correct location for the specific translators from methods on the API to internal data calls on the simulators. This has the advantage that there existing interactions between PyNN and NEURON/MOOSE, and would mean inclusion of just pynn and libneuroml packages gives an almost complete simulator independent model development and simulation environment. Can discuss this all on Monday... Padraig On 18/05/12 16:31, Mike Vella wrote: > Hi Mike, > > I think the idea is that libNeuroML will not be a > simulator-independent API, this functionality will be provided by a > separate package (current working name 'Pyramidal'). libNeuroML will > provide an object model for morphology and channel/synapse kinetics. A > lot of this will of course be separating morphforge into separate > components. There is I suppose an argument to be made for separating > morphology and kinetics into their own submodules? > > Mike > > On 14 May 2012 14:55, Michael Hull<mik...@go...> wrote: >> Hi Guys, >> Good to talk over skype, its really great to see upcoming developments. >> >> I have just been thinking about the logisitics of library maintenance. >> >> The aim is pull out the core morphology classes out of morphforge, and >> into a separate package, since we want small, well encapsulated modules. >> If we put the morphology class into libNeuroML, and this is intended as a >> simulator independent API (i.e. more than just morphologies), then we >> haven't achieved this; its the same situation that we had before, just >> replace 'morphforge' for 'libNeuroML' and we still have unnessesary >> dependancies if people just want to use the 'morphology' part of the library. >> If we are going to pull out the morphology part of the library, >> shouldn't this go in >> an entirely small, separate repository, called, for example, >> 'neuro-morph' or something >> similar, then both libNeuroML and morphforge can just import this >> repository as a >> submodule: >> >> http://git-scm.com/book/en/Git-Tools-Submodules >> >> This allows us both to use 'neuro-morph', without either side introducing >> unnessessary dependancies. Github seems to support submodules. >> (http://help.github.com/submodules/) >> >> What do you think? >> >> >> Mike >> >> ------------------------------------------------------------------------------ >> Live Security Virtual Conference >> Exclusive live event will cover all the ways today's security and >> threat landscape has changed and how IT managers can respond. Discussions >> will include endpoint security, mobile security and the latest in malware >> threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ >> _______________________________________________ >> Neuroml-python mailing list >> Neu...@li... >> https://lists.sourceforge.net/lists/listinfo/neuroml-python > ------------------------------------------------------------------------------ > Live Security Virtual Conference > Exclusive live event will cover all the ways today's security and > threat landscape has changed and how IT managers can respond. Discussions > will include endpoint security, mobile security and the latest in malware > threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ > _______________________________________________ > Neuroml-python mailing list > Neu...@li... > https://lists.sourceforge.net/lists/listinfo/neuroml-python > -- ----------------------------------------------------- Padraig Gleeson Room 321, Anatomy Building Department of Neuroscience, Physiology& Pharmacology University College London Gower Street London WC1E 6BT United Kingdom +44 207 679 3214 p.g...@uc... ----------------------------------------------------- |
|
From: Michael H. <mik...@go...> - 2012-05-25 09:07:50
|
Hello, Yes, that sounds good to me. Mike On 24 May 2012 23:23, Mike Vella <vel...@gm...> wrote: > Hi All, > > As previously discussed could we have another Skype meeting to discuss the > multi-compartmental API on Monday at 3PM UK time? > > All the best, > Mike > > > ------------------------------------------------------------------------------ > Live Security Virtual Conference > Exclusive live event will cover all the ways today's security and > threat landscape has changed and how IT managers can respond. Discussions > will include endpoint security, mobile security and the latest in malware > threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ > _______________________________________________ > Neuroml-python mailing list > Neu...@li... > https://lists.sourceforge.net/lists/listinfo/neuroml-python > > |
|
From: Mike V. <vel...@gm...> - 2012-05-24 22:23:25
|
Hi All, As previously discussed could we have another Skype meeting to discuss the multi-compartmental API on Monday at 3PM UK time? All the best, Mike |
|
From: Andrew D. <and...@un...> - 2012-05-24 11:54:28
|
Hi, I think it's a great idea to collect these files, but I think we should avoid storing large files (> 1MB) in the git repository, as it can make many operations (e.g. cloning) very slow. There are a couple of good ideas for handling this situation on this StackOverflow page: http://stackoverflow.com/questions/540535/managing-large-binary-files-with-git So perhaps we could create a repository called "example_data" or something on Github and then use git-submodule to pull it into libNeuroML? Cheers, Andrew On 23 mai 12, at 20:08, Padraig Gleeson wrote: > Hi, > > I've added a folder https://github.com/NeuralEnsemble/libNeuroML/tree/master/hdf5Examples > to the GitHub repo where anyone can put examples of HDF5 files they > use/generate for morphs/simulation results, etc. > > The 2 examples I put there which are used by neuroConstruct are just > for network structure (based on NetworkML v1.8.1) and a simple file > for storing one population's set of membrane potentials. > > Can others put any other examples of HDF5 files they use in there, > and add a note to the README (Subhasis, Michele, Stephan, you all > should have some lying around)? Andrew, can you locate any neo > compatible files containing typical simulation data? > > Anyone know any good tools apart from hdfview for browsing through > these? > > We should be able to hammer these out into a consensus format. > > Regards, > Padraig > > > > On 19/05/12 11:44, Subhasis Ray wrote: >> >> >> >> Can you post a link to your custom HDF5 format for storing simulation >> data and network structure? >> >> Stephan >> >> Hi, >> I apologize for the delayed response. Somehow I had ticked daily- >> digest instead of individual mails and received the digest for the >> whole week today. Anyways, I have not made the custom format or the >> visualization tool public yet (needs a lot of cleanup). I can give >> a description. But be warned that it is very specific to the model >> I use. I have some ideas about how it should have been, but I never >> got around to develop a clean format. >> >> For every simulation one data file and one network file is generated. >> >> The datafile has: >> /Vm [group with datasets for membrane potentials under it] >> /Vm/{cellname} [dataset for somatic Vm of cell {cellname} - I do >> not need to store the compartment info] >> /spikes [ group with datasets for spike times ] >> /spikes/{cellname} [dataset for spike-times of cell {cellname} >> /Ca [group for storing Ca concentration in cells] >> /lfp [group containing datasets for local field potentials at >> various depths] >> /stimulus [group containing datasets for stimulus applied] >> >> /network [Group] >> /network/synapse [Compound Dataset with columns: >> source: string ({source cell-name} / compartment-no) # I should >> have put the compartment id as a separate column >> dest: string ({destination cell-name} / compartment-no) >> type: string >> gbar: float # I have to maintain this as variation is introduced >> when generating network >> # the following three are redundant, and should be part of the >> individual synapse models >> tau1: float >> tau2: float >> Ek: float >> ] >> /network/hhchan [Another compound dataset to track Hodgkin-Huxley >> type channels after noise is added in network] >> # The following is again simulation info rather than model. >> /stimulus [another group to store the stimulus applied to sets of >> cells] >> /stimulus/connection [Dataset storing stimulus-source against {cell- >> name}/{compartment-no} to which it is applied] >> >> Most frequent operations are selecting datasets for subpopulations >> of cells by type (which is done using regular expression because >> the cell names start with their type name), selecting datasets for >> cells that are pre-/post- synaptic to a specific cell and finding >> correlations with various parameters (like synaptic conductance/ >> strength) in the model with the firing statistics, associating >> stimulus with response (eg PSTH). Cell names used in the network >> file are same as that in the data file allowing this kind of queries. >> >> This is rather unclean. A better way would have been to: >> 1. include the individual cell and channel prototypes in a model >> file along with the network info. >> 2. have a dataset mapping cell Ids to their prototypes and values >> of parameters that have been modified from the prototype. >> 3. organize the data file to group individual data (like Vm, Ca, >> Spike time) and population data (like LFP) separately. >> 4. store simulation experiment configuration information (stuff >> that SEDML was created for) in a more organized fashion. >> >> >> Best, >> Subhasis >> >> >> >> ------------------------------------------------------------------------------ >> Live Security Virtual Conference >> Exclusive live event will cover all the ways today's security and >> threat landscape has changed and how IT managers can respond. >> Discussions >> will include endpoint security, mobile security and the latest in >> malware >> threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ >> >> >> _______________________________________________ >> Neuroml-python mailing list >> Neu...@li... >> https://lists.sourceforge.net/lists/listinfo/neuroml-python > > ------------------------------------------------------------------------------ > Live Security Virtual Conference > Exclusive live event will cover all the ways today's security and > threat landscape has changed and how IT managers can respond. > Discussions > will include endpoint security, mobile security and the latest in > malware > threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/_______________________________________________ > Neuroml-python mailing list > Neu...@li... > https://lists.sourceforge.net/lists/listinfo/neuroml-python -- Dr Andrew Davison Unité de Neuroscience, Information et Complexité (UNIC) Institut de Neurobiologie Alfred Fessard Centre Nationale de la Recherche Scientifique 1, avenue de la Terrasse 91198 Gif sur Yvette France Tel: +33 1 69 82 34 51 http://www.andrewdavison.info/ |
|
From: Subhasis R. <ray...@gm...> - 2012-05-24 06:35:00
|
Hi, viTables (http://vitables.org/) is a python application for viewing hdf5 files. I shall look up some hdf5 file of small size from my simulations and add it to the repo. Regards, Subhasis On Wed, May 23, 2012 at 11:38 PM, Padraig Gleeson <p.g...@uc...>wrote: > Hi, > > I've added a folder > https://github.com/NeuralEnsemble/libNeuroML/tree/master/hdf5Examples to > the GitHub repo where anyone can put examples of HDF5 files they > use/generate for morphs/simulation results, etc. > > The 2 examples I put there which are used by neuroConstruct are just for > network structure (based on NetworkML v1.8.1) and a simple file for storing > one population's set of membrane potentials. > > Can others put any other examples of HDF5 files they use in there, and add > a note to the README (Subhasis, Michele, Stephan, you all should have some > lying around)? Andrew, can you locate any neo compatible files containing > typical simulation data? > > Anyone know any good tools apart from hdfview for browsing through these? > > We should be able to hammer these out into a consensus format. > > Regards, > Padraig > > > > > On 19/05/12 11:44, Subhasis Ray wrote: > > > >> Can you post a link to your custom HDF5 format for storing simulation >> data and network structure? >> >> Stephan >> > > Hi, > I apologize for the delayed response. Somehow I had ticked daily-digest > instead of individual mails and received the digest for the whole week > today. Anyways, I have not made the custom format or the visualization tool > public yet (needs a lot of cleanup). I can give a description. But be > warned that it is very specific to the model I use. I have some ideas about > how it should have been, but I never got around to develop a clean format. > > For every simulation one data file and one network file is generated. > > The datafile has: > /Vm [group with datasets for membrane potentials under it] > /Vm/{cellname} [dataset for somatic Vm of cell {cellname} - I do not need > to store the compartment info] > /spikes [ group with datasets for spike times ] > /spikes/{cellname} [dataset for spike-times of cell {cellname} > /Ca [group for storing Ca concentration in cells] > /lfp [group containing datasets for local field potentials at various > depths] > /stimulus [group containing datasets for stimulus applied] > > /network [Group] > /network/synapse [Compound Dataset with columns: > source: string ({source cell-name} / compartment-no) # I should have put > the compartment id as a separate column > dest: string ({destination cell-name} / compartment-no) > type: string > gbar: float # I have to maintain this as variation is introduced when > generating network > # the following three are redundant, and should be part of the individual > synapse models > tau1: float > tau2: float > Ek: float > ] > /network/hhchan [Another compound dataset to track Hodgkin-Huxley type > channels after noise is added in network] > # The following is again simulation info rather than model. > /stimulus [another group to store the stimulus applied to sets of cells] > /stimulus/connection [Dataset storing stimulus-source against > {cell-name}/{compartment-no} to which it is applied] > > Most frequent operations are selecting datasets for subpopulations of > cells by type (which is done using regular expression because the cell > names start with their type name), selecting datasets for cells that are > pre-/post- synaptic to a specific cell and finding correlations with > various parameters (like synaptic conductance/strength) in the model with > the firing statistics, associating stimulus with response (eg PSTH). Cell > names used in the network file are same as that in the data file allowing > this kind of queries. > > This is rather unclean. A better way would have been to: > 1. include the individual cell and channel prototypes in a model file > along with the network info. > 2. have a dataset mapping cell Ids to their prototypes and values of > parameters that have been modified from the prototype. > 3. organize the data file to group individual data (like Vm, Ca, Spike > time) and population data (like LFP) separately. > 4. store simulation experiment configuration information (stuff that SEDML > was created for) in a more organized fashion. > > > Best, > Subhasis > > > > ------------------------------------------------------------------------------ > Live Security Virtual Conference > Exclusive live event will cover all the ways today's security and > threat landscape has changed and how IT managers can respond. Discussions > will include endpoint security, mobile security and the latest in malware > threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ > > > > _______________________________________________ > Neuroml-python mailing lis...@li...https://lists.sourceforge.net/lists/listinfo/neuroml-python > > > > > ------------------------------------------------------------------------------ > Live Security Virtual Conference > Exclusive live event will cover all the ways today's security and > threat landscape has changed and how IT managers can respond. Discussions > will include endpoint security, mobile security and the latest in malware > threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ > _______________________________________________ > Neuroml-python mailing list > Neu...@li... > https://lists.sourceforge.net/lists/listinfo/neuroml-python > > |
|
From: Padraig G. <p.g...@uc...> - 2012-05-23 18:08:52
|
Hi, I've added a folder https://github.com/NeuralEnsemble/libNeuroML/tree/master/hdf5Examples to the GitHub repo where anyone can put examples of HDF5 files they use/generate for morphs/simulation results, etc. The 2 examples I put there which are used by neuroConstruct are just for network structure (based on NetworkML v1.8.1) and a simple file for storing one population's set of membrane potentials. Can others put any other examples of HDF5 files they use in there, and add a note to the README (Subhasis, Michele, Stephan, you all should have some lying around)? Andrew, can you locate any neo compatible files containing typical simulation data? Anyone know any good tools apart from hdfview for browsing through these? We should be able to hammer these out into a consensus format. Regards, Padraig On 19/05/12 11:44, Subhasis Ray wrote: > > > Can you post a link to your custom HDF5 format for storing simulation > data and network structure? > > Stephan > > > Hi, > I apologize for the delayed response. Somehow I had ticked > daily-digest instead of individual mails and received the digest for > the whole week today. Anyways, I have not made the custom format or > the visualization tool public yet (needs a lot of cleanup). I can give > a description. But be warned that it is very specific to the model I > use. I have some ideas about how it should have been, but I never got > around to develop a clean format. > > For every simulation one data file and one network file is generated. > > The datafile has: > /Vm [group with datasets for membrane potentials under it] > /Vm/{cellname} [dataset for somatic Vm of cell {cellname} - I do not > need to store the compartment info] > /spikes [ group with datasets for spike times ] > /spikes/{cellname} [dataset for spike-times of cell {cellname} > /Ca [group for storing Ca concentration in cells] > /lfp [group containing datasets for local field potentials at various > depths] > /stimulus [group containing datasets for stimulus applied] > > /network [Group] > /network/synapse [Compound Dataset with columns: > source: string ({source cell-name} / compartment-no) # I should have > put the compartment id as a separate column > dest: string ({destination cell-name} / compartment-no) > type: string > gbar: float # I have to maintain this as variation is introduced when > generating network > # the following three are redundant, and should be part of the > individual synapse models > tau1: float > tau2: float > Ek: float > ] > /network/hhchan [Another compound dataset to track Hodgkin-Huxley type > channels after noise is added in network] > # The following is again simulation info rather than model. > /stimulus [another group to store the stimulus applied to sets of cells] > /stimulus/connection [Dataset storing stimulus-source against > {cell-name}/{compartment-no} to which it is applied] > > Most frequent operations are selecting datasets for subpopulations of > cells by type (which is done using regular expression because the cell > names start with their type name), selecting datasets for cells that > are pre-/post- synaptic to a specific cell and finding correlations > with various parameters (like synaptic conductance/strength) in the > model with the firing statistics, associating stimulus with response > (eg PSTH). Cell names used in the network file are same as that in > the data file allowing this kind of queries. > > This is rather unclean. A better way would have been to: > 1. include the individual cell and channel prototypes in a model file > along with the network info. > 2. have a dataset mapping cell Ids to their prototypes and values of > parameters that have been modified from the prototype. > 3. organize the data file to group individual data (like Vm, Ca, Spike > time) and population data (like LFP) separately. > 4. store simulation experiment configuration information (stuff that > SEDML was created for) in a more organized fashion. > > > Best, > Subhasis > > > > ------------------------------------------------------------------------------ > Live Security Virtual Conference > Exclusive live event will cover all the ways today's security and > threat landscape has changed and how IT managers can respond. Discussions > will include endpoint security, mobile security and the latest in malware > threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ > > > _______________________________________________ > Neuroml-python mailing list > Neu...@li... > https://lists.sourceforge.net/lists/listinfo/neuroml-python |
|
From: Andrew D. <and...@un...> - 2012-05-23 11:30:25
|
On 22 mai 12, at 15:50, Padraig Gleeson wrote: > This raises a good question: which version of NeuroML should the API > initially target? All else being equal, I think we should target v2.0. A reason for targeting v1.8.1 intially would be if there are not enough existing models in 2.0 format. > I'd suggest having the internal data model as close to NML v2.0 as > possible, and (eventually) the API can natively support import and > export of NML v1.8.1, potentially by using an API automatically > generated from the v1.x Schemas. > > Padraig |
|
From: Padraig G. <p.g...@uc...> - 2012-05-22 13:50:35
|
Hi all, In response to a much earlier email from Stephan, an outline of the structure of NeuroML 2 is available here: http://www.neuroml.org/NeuroML2CoreTypes/Cells.html. This raises a good question: which version of NeuroML should the API initially target? I'd suggest having the internal data model as close to NML v2.0 as possible, and (eventually) the API can natively support import and export of NML v1.8.1, potentially by using an API automatically generated from the v1.x Schemas. Padraig -------- Original Message -------- Subject: Re: Python libNeuroML Date: Fri, 23 Mar 2012 17:18:40 -0400 From: Stephan Gerhard <con...@un...> To: P Gleeson <p.g...@uc...> CC: Mike Vella <mv...@ca...>, <and...@un...>, <mat...@eb...>, <phi...@g-...>, Michael Hull <mik...@go...>, <r.j...@ne...>, <san...@as...>, <ko...@ki...> Padraig, Do you have a nice hierarchical diagram of the NeuroMLv2, it is a bit hard to read just from text? Thanks, Stephan |
|
From: Subhasis R. <ray...@gm...> - 2012-05-19 10:44:35
|
>
> Can you post a link to your custom HDF5 format for storing simulation
> data and network structure?
>
> Stephan
>
Hi,
I apologize for the delayed response. Somehow I had ticked daily-digest
instead of individual mails and received the digest for the whole week
today. Anyways, I have not made the custom format or the visualization tool
public yet (needs a lot of cleanup). I can give a description. But be
warned that it is very specific to the model I use. I have some ideas about
how it should have been, but I never got around to develop a clean format.
For every simulation one data file and one network file is generated.
The datafile has:
/Vm [group with datasets for membrane potentials under it]
/Vm/{cellname} [dataset for somatic Vm of cell {cellname} - I do not need
to store the compartment info]
/spikes [ group with datasets for spike times ]
/spikes/{cellname} [dataset for spike-times of cell {cellname}
/Ca [group for storing Ca concentration in cells]
/lfp [group containing datasets for local field potentials at various
depths]
/stimulus [group containing datasets for stimulus applied]
/network [Group]
/network/synapse [Compound Dataset with columns:
source: string ({source cell-name} / compartment-no) # I should have put
the compartment id as a separate column
dest: string ({destination cell-name} / compartment-no)
type: string
gbar: float # I have to maintain this as variation is introduced when
generating network
# the following three are redundant, and should be part of the individual
synapse models
tau1: float
tau2: float
Ek: float
]
/network/hhchan [Another compound dataset to track Hodgkin-Huxley type
channels after noise is added in network]
# The following is again simulation info rather than model.
/stimulus [another group to store the stimulus applied to sets of cells]
/stimulus/connection [Dataset storing stimulus-source against
{cell-name}/{compartment-no} to which it is applied]
Most frequent operations are selecting datasets for subpopulations of cells
by type (which is done using regular expression because the cell names
start with their type name), selecting datasets for cells that are
pre-/post- synaptic to a specific cell and finding correlations with
various parameters (like synaptic conductance/strength) in the model with
the firing statistics, associating stimulus with response (eg PSTH). Cell
names used in the network file are same as that in the data file allowing
this kind of queries.
This is rather unclean. A better way would have been to:
1. include the individual cell and channel prototypes in a model file along
with the network info.
2. have a dataset mapping cell Ids to their prototypes and values of
parameters that have been modified from the prototype.
3. organize the data file to group individual data (like Vm, Ca, Spike
time) and population data (like LFP) separately.
4. store simulation experiment configuration information (stuff that SEDML
was created for) in a more organized fashion.
Best,
Subhasis
|
|
From: Mike V. <vel...@gm...> - 2012-05-18 15:31:47
|
Hi Mike, I think the idea is that libNeuroML will not be a simulator-independent API, this functionality will be provided by a separate package (current working name 'Pyramidal'). libNeuroML will provide an object model for morphology and channel/synapse kinetics. A lot of this will of course be separating morphforge into separate components. There is I suppose an argument to be made for separating morphology and kinetics into their own submodules? Mike On 14 May 2012 14:55, Michael Hull <mik...@go...> wrote: > Hi Guys, > Good to talk over skype, its really great to see upcoming developments. > > I have just been thinking about the logisitics of library maintenance. > > The aim is pull out the core morphology classes out of morphforge, and > into a separate package, since we want small, well encapsulated modules. > If we put the morphology class into libNeuroML, and this is intended as a > simulator independent API (i.e. more than just morphologies), then we > haven't achieved this; its the same situation that we had before, just > replace 'morphforge' for 'libNeuroML' and we still have unnessesary > dependancies if people just want to use the 'morphology' part of the library. > If we are going to pull out the morphology part of the library, > shouldn't this go in > an entirely small, separate repository, called, for example, > 'neuro-morph' or something > similar, then both libNeuroML and morphforge can just import this > repository as a > submodule: > > http://git-scm.com/book/en/Git-Tools-Submodules > > This allows us both to use 'neuro-morph', without either side introducing > unnessessary dependancies. Github seems to support submodules. > (http://help.github.com/submodules/) > > What do you think? > > > Mike > > ------------------------------------------------------------------------------ > Live Security Virtual Conference > Exclusive live event will cover all the ways today's security and > threat landscape has changed and how IT managers can respond. Discussions > will include endpoint security, mobile security and the latest in malware > threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ > _______________________________________________ > Neuroml-python mailing list > Neu...@li... > https://lists.sourceforge.net/lists/listinfo/neuroml-python |
|
From: Michele M. <mat...@gm...> - 2012-05-16 10:41:08
|
Hello people, given the fact we are talking about HDF structure to save simulations results, I want to point out the HDF we use in Neuronvisio: http://neuronvisio.org/storage.html#hdf-structure Basically it has a branch, called geometry, where the leaf consist in a string of NeuroML (version 1.8 for now) and a branch called results, which branches into a VecRef and can be expanded to user need to other Ref, that's what the GenericRef is about. The VecRef branches into section->variable_name, with variable_name being a numpy array. Just one more way which we could take inspiration from. P.S.: we decided to give Neuronvisio a proper home, with a more classic neuronvisio.org. Please use this one if you need to reference the website. Thanks. Cheers, Michele. On Tue, May 15, 2012 at 6:06 PM, Stephan Gerhard <uni...@gm...> wrote: > Hi, > > Can you post a link to your custom HDF5 format for storing simulation > data and network structure? > > Stephan > > ------------------------------------------------------------------------------ > Live Security Virtual Conference > Exclusive live event will cover all the ways today's security and > threat landscape has changed and how IT managers can respond. Discussions > will include endpoint security, mobile security and the latest in malware > threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ > _______________________________________________ > Neuroml-python mailing list > Neu...@li... > https://lists.sourceforge.net/lists/listinfo/neuroml-python |
|
From: Stephan G. <uni...@gm...> - 2012-05-15 17:06:33
|
Hi, Can you post a link to your custom HDF5 format for storing simulation data and network structure? Stephan |
|
From: Subhasis R. <ray...@gm...> - 2012-05-15 11:10:08
|
Hi, While I could not participate in skype meeting, Chaitanya updated me regarding it. Although I weaved a custom HDF5 format for simulation data and network models in my own work couple of years back and it works fine for me, I have been thinking about document oriented approach vs database oriented approach for some time. While the serialization of NeuroML into HDF5 files is a most welcome step, I would like to point out that we can take this opportunity to go one step further and design a database oriented approach to storing models and data and linking them. This is only a small step of generalization when thinking about the tables/arrays in HDF5, but will open up the possibilities further. My impression of NeuroML and other neuronal model description languages has been that they are most commonly used for hand coded models with very low entropy (that is, even if variability is introduced, it is introduced using rules that can be described precisely). For example, one takes simplified prototype cell models and makes 1000 replica of it and then modifies some property systematically (possibly using a random number generator) and connects them up according to some connection probabilities. This works fine in studies where theory is of main importance. But with data driven projects like the Whole Brain Catalog coming up and many groups working towards creating detailed maps of the brain at various levels, I sense a need for database oriented approach. If we want to bridge the gap between experimental data and computational models, we need to consider is handling large amounts of data generated from experiments. While MorphML has been used to convert neuronal morphology data dumped by machines, I suspect that in future it will become more important to store detailed data on network structures (as well as biophysical properties) and interface that data seamlessly with simulation environments (In an ideal situation a researcher should be able to make simple queries and updates in a model to see how changing parameters change the behaviour: for example, "what is expected to happen if I change my ACSF to have 3 mM KCl in stead of 1 mM", without killing another rat or spending a week to get significant numbers). With that kind of data, XML will be hard to scale as parsing becomes more and more memory and time consuming with file size. On top of that, if things get query oriented, even HDF5 may become unsuited for such operations and interfacing with a database will become necessary. On the computational side, when analyzing data from simulations of large network models, I find myself going back and forth between the model definition and the data file. While custom HDF5 format with python functions to process it works fine for a one-man-project, a more general approach is desirable. Some of my most used functions are just for querying the network structure to find out which array (of time series data) in the data file is of interest. Another common usage is extracting various statistics and graph-theoretic information from the network structure itself based on constraints like cell type or synapse type. This kind of work flow asks for a database oriented approach rather than a document-based approach. Other use of model/structure in combination with time series or other data is for visualizers (Neuronvisio, Moogli). In their most general form, these tools should be usable for both experimental as well as simulated data. Here also a database oriented approach will be easier to map user's requests to model components and the corresponding data. Coordination with groups working on formats for electrophysiological data (neo for example) to have a unified format for data along with standards to link the model to the data will be useful in this respect. Finally, my first impression of the current form of neuroML 2 is that it is well suited for such serialization. One can go through step by step normalization and design a set of database tables that map to the neuroML structures. This can serve for designing the HDF5 format as well. As I mentioned before, we are using a custom HDF5 format for storing simulation data as well as network structure. We shall be glad to merge these with a carefully designed standard format in the future. I am eager to know what the community thinks. Regards, Subhasis |
|
From: Stephan G. <uni...@gm...> - 2012-05-15 07:54:07
|
Hi, I did not point out this yesterday in the discussion, but it could be of interest to Mike Vella for inspiration regarding defining the object model for morphologies. We have an object model for neural circuits in CATMAID (postgres database schema), we export that to NeuroHDF for people wanting to do postprocessing in Matlab/Python or visualization. The HDF5 datastructure is here: https://github.com/acardona/CATMAID/blob/cmw-integration/django/applications/vncbrowser/views/neurohdf.py#L240 I wrote a simple set of wrapper classes that can load such an exported NeuroHDF file and exposes a simple object model with convenience functions: https://github.com/unidesigner/microcircuit/blob/master/microcircuit/circuit.py And I can use the same schema to visualize circuits in fos: https://github.com/fos/fos/blob/master/examples/microcircuit_neurohdf.py I would appreciate it if you guys can come up with an object model that is at least as expressive as the ones above, so that it will be straightforward to write a mapper. Stephan On Mon, May 14, 2012 at 3:55 PM, Michael Hull <mik...@go...> wrote: > Hi Guys, > Good to talk over skype, its really great to see upcoming developments. > > I have just been thinking about the logisitics of library maintenance. > > The aim is pull out the core morphology classes out of morphforge, and > into a separate package, since we want small, well encapsulated modules. > If we put the morphology class into libNeuroML, and this is intended as a > simulator independent API (i.e. more than just morphologies), then we > haven't achieved this; its the same situation that we had before, just > replace 'morphforge' for 'libNeuroML' and we still have unnessesary > dependancies if people just want to use the 'morphology' part of the library. > If we are going to pull out the morphology part of the library, > shouldn't this go in > an entirely small, separate repository, called, for example, > 'neuro-morph' or something > similar, then both libNeuroML and morphforge can just import this > repository as a > submodule: > > http://git-scm.com/book/en/Git-Tools-Submodules > > This allows us both to use 'neuro-morph', without either side introducing > unnessessary dependancies. Github seems to support submodules. > (http://help.github.com/submodules/) > > What do you think? > > > Mike > > ------------------------------------------------------------------------------ > Live Security Virtual Conference > Exclusive live event will cover all the ways today's security and > threat landscape has changed and how IT managers can respond. Discussions > will include endpoint security, mobile security and the latest in malware > threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ > _______________________________________________ > Neuroml-python mailing list > Neu...@li... > https://lists.sourceforge.net/lists/listinfo/neuroml-python |
|
From: <cha...@nc...> - 2012-05-15 04:33:49
|
Hi Mike, Yes I am aware of Neuronvisio. The only difference from my understanding is that, Neuronvisio is closely linked to Neuron and Moogli is not (not even to Moose). Technically, Moogli uses OpenGL and Neuronvisio uses Mayavi. I did not mean to duplicate efforts, but since I already wrote the basic code for Moosegui (which is Moose equivalent of Neuronvisio), I just spent some time to make it independent. Infact, to use HDF format that I have adopted is based on Neurovisio's ideas among other reasons. The exact equivalent of Moogli, I believe is Neuranim: https://sourceforge.net/projects/neuranim/ Some cool features in Moogli include, 'direction' of movies (keyframes etc.), and also has the capabilities to do stereoscopic rendering (we have tested this). The idea is to make it like PyMol. It would be great if any of you would like to collaborate on this project. Cheers! Chaitanya > Hi Chaitanya, > > Have you checked out neuronvisio? http://michelemattioni.me/neuronvisio/ > > Mike > > On 14 May 2012 15:52, <cha...@nc...> wrote: > >> Hey guys, >> >> I am Chaitanya from NCBS. (We were in call just a while ago) >> >> I have been developing Moogli for some time now. I thought you could >> give >> your comments on it. It is a simulator independent, offline, opengl >> based >> visualizer for neural simulations. This was initially part of the >> Moosegui >> development, that has branched out to be independent. >> >> Currently supported formats for morphology inspection are, .xml >> (MorphML) >> and .csv. For data visualization, I employ a very primitive hdf5 format >> for saving time-series data. The specific details of which are >> elaborated >> in the webpage below. >> >> Soon, moogli shall also support network level visualizations. >> >> It has been suggested by Upi, that I incorporate support for neo which >> would enable moogli to visualize electrophys data as well. Also, it is >> only recently that I have come to know about the efforts towards >> neuroHDF. >> >> Here is a primitive webpage: http://moose.ncbs.res.in/moogli/ >> It is in a git repo: https://github.com/ccluri/Moogli >> >> Cheers! >> Chaitanya >> >> >> >> ------------------------------------------------------------------------------ >> Live Security Virtual Conference >> Exclusive live event will cover all the ways today's security and >> threat landscape has changed and how IT managers can respond. >> Discussions >> will include endpoint security, mobile security and the latest in >> malware >> threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ >> _______________________________________________ >> Neuroml-python mailing list >> Neu...@li... >> https://lists.sourceforge.net/lists/listinfo/neuroml-python >> > |
|
From: Mike V. <vel...@gm...> - 2012-05-14 16:23:01
|
Hi Chaitanya, Have you checked out neuronvisio? http://michelemattioni.me/neuronvisio/ Mike On 14 May 2012 15:52, <cha...@nc...> wrote: > Hey guys, > > I am Chaitanya from NCBS. (We were in call just a while ago) > > I have been developing Moogli for some time now. I thought you could give > your comments on it. It is a simulator independent, offline, opengl based > visualizer for neural simulations. This was initially part of the Moosegui > development, that has branched out to be independent. > > Currently supported formats for morphology inspection are, .xml (MorphML) > and .csv. For data visualization, I employ a very primitive hdf5 format > for saving time-series data. The specific details of which are elaborated > in the webpage below. > > Soon, moogli shall also support network level visualizations. > > It has been suggested by Upi, that I incorporate support for neo which > would enable moogli to visualize electrophys data as well. Also, it is > only recently that I have come to know about the efforts towards neuroHDF. > > Here is a primitive webpage: http://moose.ncbs.res.in/moogli/ > It is in a git repo: https://github.com/ccluri/Moogli > > Cheers! > Chaitanya > > > > ------------------------------------------------------------------------------ > Live Security Virtual Conference > Exclusive live event will cover all the ways today's security and > threat landscape has changed and how IT managers can respond. Discussions > will include endpoint security, mobile security and the latest in malware > threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ > _______________________________________________ > Neuroml-python mailing list > Neu...@li... > https://lists.sourceforge.net/lists/listinfo/neuroml-python > |
|
From: <cha...@nc...> - 2012-05-14 15:22:44
|
Hey guys, I am Chaitanya from NCBS. (We were in call just a while ago) I have been developing Moogli for some time now. I thought you could give your comments on it. It is a simulator independent, offline, opengl based visualizer for neural simulations. This was initially part of the Moosegui development, that has branched out to be independent. Currently supported formats for morphology inspection are, .xml (MorphML) and .csv. For data visualization, I employ a very primitive hdf5 format for saving time-series data. The specific details of which are elaborated in the webpage below. Soon, moogli shall also support network level visualizations. It has been suggested by Upi, that I incorporate support for neo which would enable moogli to visualize electrophys data as well. Also, it is only recently that I have come to know about the efforts towards neuroHDF. Here is a primitive webpage: http://moose.ncbs.res.in/moogli/ It is in a git repo: https://github.com/ccluri/Moogli Cheers! Chaitanya |
1 message has been excluded from this view by a project administrator.
