Acknowledgement

multi-electroderecordings

revisionfromFebruary11,2006

1SomeSmallPrint

Manyofthedesignationsusedbymanufacturersandsellerstodistinguishtheirpro-ductsareclaimedastrademarks.Wherethosedesignationsappearinthismanual,andwewereawareofatrademarkclaim,thedesignationshavebeensetincapsorinitialcaps.Whileeveryprecautionhasbeentakeninthepreparationofthismanual,weassumenoresponsibilityforerrorsoromissions,ordamagesfromtheuseoftheinformationcontainedneurcherein.

V.Berenguer,2005anddistributed󰀁

cALCis󰀁

underGPLversion21.ItsmanualisM.Bongard,V.Berenguer,2005,andlicensedunderaCreativeCommonsLicense2.TheprogramisbasedTrolltech’sQt3.3C++framework3andthelibrariesQWT4,aswellasQWT3D5forthegraphicaluserinterface.Severalimplementedfunctionsofthepro-gramaretakenfromtheTISEANpackage6.ThecalculationoffouriertransformationsusesroutinesfromtheFFTW3library7.TheunsupervisedclusteranalysisofthePSTHdataisbasedonKlustakwik8.DatabasefunctionalityisprovidedbyutilizingMySQL9.

Acknowledgement

neurALCVersion1.0wasdevelopedduringjuly-october2005inthegroupofProf.Dr.E.FernandezattheinstituteofbioengineeringoftheuniversityMiguelHernandez,Alicante(Spain).WeliketothankeveryonewhocontributedtothemakingofneurALC..

ConventionsWithinThisManual

•Referencesaresetusingtoacomponentstypewriterofface.theGUI,commandsorotherfunctionsofneurALC•Aspectsmarginalweexclamationliketoemphasizeicon(likearethesetoneusingshownaonromantheleft).typefaceandmarkedbya•ThelayoutofthemanualforneurALCisintendedfortwo-sidedprinting.

•Taextualhierarchicaldescriptionimplementation.

ofmenusorcommandstructuresuses→-symbolstoillustrate12http://www.gnu.org

3http://creativecommons.org/licenses/by-nc-sa/2.0/

4

http://www.trolltech.com5

http://qwt.sourceforge.net6

http://qwt3dplot.sourceforge.net7

http://www.mpiks-dresden.mpg.de/∼tisean8

http://www.fftw.org9

http://klustakwik.sourceforge.nethttp://www.mysql.com

2

2Preface

neurALC-isaprogramintendedforworkingandanalyzingneuronalmulti-electroderecordings.Inthecurrentversionitofferssomefunctionalitytorevealandanalyzesomeoftheinformationcontainedinsuchrecordingsanditisfurtherintendedasakindof“catalyst”forthedevelopmentofafreeavailablecross-platformprogramfortheanaly-sisofelectrophysiologicalrecordings,whichisdistributedundertheGPL(GNUPublicLicence2.0).

Thismanualisintendedasshortdescriptionoftheprovidedfunctionalityanditstech-nicalbackground,aswellasofferingasometutorialandprogramminginformation.

TheName...

Well,theprogram,orbetteritsnamecomealongway.Forasmallgroupofpeoplethesoftwareisalsoknownastheprogramformerlycalled“WAND”,“PROBE”,or“Genie”.ThesenamesweredismissedafterwediscoveredthatforexampletheWideAreaNet-workDaemonwassimplytherefirst,andwell,therearesometrademarksoneditablelubricants,etc.whichusessomeofthenameswethoughtofinthebeginning-inshort:neurALCiscomposedfromtwosyllables-neurderivatedfromthefactthatthepro-gramworkswithneuronaldata,andALC,whichistheabbreviationusedininternationalaviationfortheAlicanteairport.

3Installation

neurALCusesTrolltech’sQt3.3C++frameworkandthelibrariesQWTandQWT3Dforthegraphicaluserinterface.SeveralimplementedfunctionsoftheprogramaretakenfromtheTISEANpackage.ThecalculationoffouriertransformationsusesroutinesfromtheFFTW3library.DatabasefunctionalityisprovidedbyutilizingMySQL.Ingeneraltermstheprogramshouldthereforerunonanyplatformonwhichthesepackagesandlibrariesareavailable.

AsofthiswritingneurALCisprovidedinfourdifferent\"flavours\":binaryinstallersforLinux,MacOSX10.3(orhigher),Windows95(orhigher)andasourcedistribution(ifyoumanagetocompileabinaryonanyotherplatform,pleasecontactus.We’rehappytointegrateanddistributemoreinstallersontheneurALCwebsite).Downloadthepackageofyourchoicefromhttp://neuralc.sourceforge.net.

3.1Linux

ThebinaryinstallerforLinuxonx86processorcomesasagnu-zippedtararchive.Afterdownloadfromsourceforgeanddearchivingneuralc.tgz,runtheinstall-neuralc.shshellscriptfromwithinbash.PrerequisitesforusingthisbinaryareaninstallationofTrolltech’sQtframeworkversion3.x(KDEversion3comeswithlibraries),theQWTver-sion0.4.2,QWT3Dversion0.2.6andtheFFTWlibrarypackage.Tousethedatabase

3

functionalityanadditionalinstallationofMySQL(version3orhigher)isneeded.De-pendingontheLinux-distributionyouuse,differentwaysofinstallingMySQLonyoursystemexist.Refertohttp://www.mysql.comforadditionalinformation.

3.2MacOSX

TheMacOSXinstallerrequiresMacOSX10.3orhigher.Thebinaryinstallerpackageincludestheprogramandallrequiredlibraries(Qt3.3.4incl.supportforMySQL,qwt0.42,qwt3d0.25,fftw3)forMacOSX.Downloadthefileneuralc1.0-installer.zip,dearchiveanddoubleclicktheicontoinstall.TousethedatabasefunctionalityanadditionalinstallationofMySQL(version3orhigher)isneeded.Visithttp://www.mysql.comanddownloadtheMySQL-installerpackageforMacOSXofyourchoice.

3.3Windows

ThebinaryinstallerforMicrosoftWindowsisbasedonNSIS10andinstallstheprogram,examplefiles,manual,sourcesandallrequiredlibraries.DownloadneurALC-win32.exefromsourceforge,unzipitandclickontheinstallericon.TheprogramisinstalledinitsowndirectoryinC:\\Programs\\neurALCandregisteredasentrywithintheWindowsSTART-buttonmenu.Allrequiredlibrariesareinstalledinthewindowssystemdirec-tory.TousethedatabasefunctionalityanadditionalinstallationofMySQL(version3orhigher)isneeded.Visithttp://www.mysql.comanddownloadtheMySQL-installerpackageforWindowsofyourchoice.

3.4Compilation

TheprogramiswritteninC++.TocompileunderyourflavorofMacOSX,*nixorWin-dows,firstdownloadandinstalltherequiredlibrariesQWT,QWT3DandFFTW,aswellasMySQL.ThendownloadtheneurALCsourcecode-archive.Extractallfilestoasingledirectory,openashell,changetothedirectoryandtypeqmake,thenmake.Thatshouldbeallyouneedtodo.UnderMacOSXand*nixgccisused;tousetheprovided.pro-fileunderWindows,aninstallationofthemingw-packagewithsuchfunc-tionality,isrecommended.Ifyourunatthisstageintoanerror,generateaprojectfileusingTrolltech’sqmakeorchangethesuppliedneuralc.proaccordingtoyourQtinstallationandC++compilerrequirements.Asofthiswritingthenon-commercialver-sionoftheQt3.2frameworkforWindowsisavailableonlyasaCD-Romadd-onofthebookBlanchette,Summerfield:“C++GUIprogrammingwithQt3”,ISBN0131240722(whichoffersingeneralaveryniceintroductiontoQtprogramming).TodownloadtheQPL/GPLversionforotherplatformsvisittheTrolltechwebpage.

thePleasebeawarethatQt3.xisneitherbinary-norsourcecode-compatibletoQt3sourceisimplicitlycodeofneeded.

neurALC,oruseoneoftheprovidedbinaries,aninstallationQt4.ofToTrolltech’scompile10

http://nsis.sourceforge.net

4

4Usage&Functionality

4.1TheneurALCGUI

Doubleclicktheprogramicon(shownontheright)torunneurALC.Aftertheprogramstart,openadatafilewithaneuronalrecordinginNEV-format(atthemomentneurALCcanonlyopenrecordingfilesintheNeuronalEventFormatVersion2).Yourscreenshouldlooksimilartotheoneshowninthefigure1(allscreenshotsinthismanualweretakenunderMacOSX;Trolltech’sQtprovidesnativelook&feelforallsupportedplatforms-dependingontheoperatingsystemyouuse,thegraphicaluserexperiencemightslightlydiffer).neurALC’suserinterfaceisdesignedinfive(atleastwethink)logicalsections.Themaingraphicaluserinterfacerepresentsthehubtoallavailablevisualizationsandprogramparameter.Itprovidesaccessviafourtabs:4.1.1ThePopulation-Tab

Asdefaultafterstart-up,neurALC’sPopulation-tabbecomesselected/active.Itissub-dividedinaninformation(figure1←1)andavisualization(figure1←2)part.Informationregardingthenumberofelectrodes(activechannels),date(creationdate),numberofrecordedneuronalspikes(spikes),samplerate(inHz),andthesortingstatusofunits-(units:sorted/unsorted)orTRC(classes:sorted/unsorted)fortheopenedexperimentalfileisprovided(asfarasitisfiledwithinthefile).Additionallyanexisting

Figure1:neurALCGUI:Population-tab

5

commentwithupto256charactersisdisplayedandcanbeeditedorcreatedviatheprovidedtextentryfield.Thelowerpartofthetab(figure1←2)isusedtovisualizethecurrentdatasetoranalysisresult-onthelevelpopulationorexperiments.Thatimpliesthattheresultofanyselectedcalculationorvisualizationwillusealldata-therecorded“population’-’providedbytheopenedexperimentaldatafile.Thesliderwidgetatthebuttonofthewindowallows,ifapplicable,tointeractivelyadjustspecificvisualizationoranalysisparameter.4.1.2TheElectrodes-Tab

SelectingtheElectrodes-tab(Figure2)allowstheusertoaccess,analyzeandvisu-alizetheinformationprovidedintheactualopenedexperimentalfileonelectrode-,(ifpreviouslysortedandcomprisedintheNEV-file)neuronalunit-,and,ifthisanalysiswasperformed,temporalresponseclass(TRC)-level.LikeforthePopulation-tab(p.5),

Figure2:neurALCGUI:Electrodes-tab

thegraphicaluserinterfaceissubdividedin2parts.One(figure2←1)whichprovidesaccesstothesingleelectrodes(Channel),neuronalunits(Unit;onlyaccessibleifthisinformationiscontainedintheactualopenedexperimentalfile),andtemporalresponseclassinformation(Class;onlyaccessibleifpreviouslycalculatedandassigned).Thelower,secondpartoftheElectrodes-tabservesasdisplayforthevisualizationofthedataoranalysisresults(figure2←2)part.Thesliderwidgetatthebuttonofthewindowallows,ifapplicable,tointeractivelyadjustspecificvisualizationoranalysisparameter.

6

4.1.3The3D-Tab

neururethepartHome3ALC←1)offersafewvisualizationinthreedimensions.TheOption-pop-upmenu(fig-ofthe-buttoninthewindow.resetsupperUsingthepartviewallowstheofmousethetoselectactualoneoftheprovidedvisualization.Clickingitispossible3D-visualizationtorotate,(figurezoom,3←or2)navigateintheloweroth-

Figure3:neurALCGUI:3D-tab

erwiseneurwithinthethree-dimensionalvisualization.Appendix-8.1includesatablewithtion.ALC’sadjustThespecificsliderkeyboardvisualizationwidgetshortatthecutsorbuttonandaanalysisofdescriptiontheofthepossiblemouse-basednaviga-parameter.

windowallows,ifapplicable,tointeractivelybeWhetherinneurALC’sPopulation-,Electrodes-or3D-tab:anyPrint-dialogue.saved-dependingontheconfigurationonthesystemwhichisvisualizationused-intoadisplayedfileusingcansaveanyplottoViaavector-basedthisOS-providedPDF-file.

functionalityitisforinstanceunderMacOSXpossibletheto7

4.1.4TheOptions-Tab

Thistabprovidesaccesstoallanalysisandvisualizationparameter.Theadjustableparameterarearrangedin7different,tabbedviewsrelatedto:theTriggerinformationwithintheopenedexperimentaldatafile,Binning-ISI(interspikeinterval)-Frequency,Correlation,Delay,Classification,Selection,and“elUltimo”.Anytuninginoneoftheoptionsisappliedtoallrelated,furthercalculationsorvisualizationduringtheactualprogramsession.

Figure4:neurALCGUI:Options-Tab

TriggerThistaballowstoselectandadjustthestimulusrelatedtimecode(“trigger”information)whichisfiledwithintheactualopenedexperimentaldatafile(orusedforthe“automatic”analysisofadirectoryoffiles).TheNEV2-formatdefinition11allowstostore2differentdatatypeswiththisinformation:“analog”and“digital”datapackets.Upto5differentlytaggedanalog(Analog1,Analog2,etc.),andonedigitalpacketidentifier(taggedasDigital)canbestoredwithinthefiles.Bothtypescanbeselected,usedasbaseforvisualization,ortheirtimestampinformationeditedfromwithinthiswindow.UsetheStart-pop-upmenu(figure5←a)toselectthepacket-labelwhichmarksthebeginningofastimulus.Ifthesameanalogordigitaldatapacketisusedtomarkstartandendofastimulation,adjusttheincrementbetweensuccessivestimulationsviathe

11

http://cybernetics.com/NEVspc20.pdf

8

textentryfieldinthesameline.Ifadifferentdatapacketisusedtomarktheendofthestimulation,choosethecorrespondingdatapacket-labelfromtheEnd-pop-pup.

Figure5:Options-Tab:Trigger

Informationofthenumberoftheselectedtriggereventsisprovidedinfigure5←b.Thesamplerateatwhichtheseinformationwasacquiredduringtheexperimentinfigure5←c.Figure5←ddisplaysalistofthetimestampsoftheselectedtriggerpacket/s(timestampsaredisplayedwiththeirsamplenumbersavedduringtheacquisitionoftheactualopenedexperiment).Upto32756differentofthesetimestampscanbedirectlyeditedfromwithinthistabbyclickingthetheEditTrigger-button(figure5←e).Binning,ISI,FrequencyBinning,interspikeinterval(ISI),andfrequencyrelatedparametercanbeaccessedfromthiswindow.

Figure6:Options-Tab:Binning,ISI,Frequency

Usethewidgetfigure6←atoadjustthebinsizetomeettherequirementsneeded.AdjusttherangewithintheinterspikeintervalsareextractedanddisplayedfromtheactualopenedfileusingtheMinor-andMajor-widgetinthewindowpartmarkedas

9

←b.Thefrequencyrangeforinstantfiringratecalculationscanbeadjustedwiththewidgetsinfigure6←c.Thewindowlengthandtheoverlapwhichisused,ifaspectrogramcalculationiscarriedout,canbeadjustedviathewidgetsmarkedasfigure6←d.

CorrelationThiswindowallowsthedefinitionofallparameterforcorrelation-relatedcalculations.UsetheWindow-textfieldmarkedinfigure7with←atoadjustthesym-metricaltimewindowinwhichthecorrelationfunctionisdetermined.SetfromtheChannel-pop-up(7←b)thereferenceelectrodeand,ifexisting,thereferenceneuronalUnit(figure7←c)ortemporalresponseClass(figure7←d)forthecorrelationcalcu-lation.ChoosingAllfromtheChannel-pop-up(7←b)meansthatthecorrelationoftheselectedelectrodesiscalculatedagainstthepopulationactivity(cumulativeactivityofallrecordedeventsexclusivetheelectrode/unitforwhichthecorrelationfunctionisdetermined.)

Figure7:Options-Tab:Correlation

DelayAdjustthetimelag(Delay,figure8←a)andDimension(figure8←b)numberwhichareusedfordelayembeddings.Theseparameterinfluencethephasespacereconstructionoftheopenedorselectedexperimentaldataset.Themethodistaken

Figure8:Options-Tab:Delay

fromtheTISEANpackageandusedinafewanalysisfunctionsofneurALC.

10

classificationneurALCofferstheoptiontoclassifywithinsingleormultipleexperi-mentsregisteredspikesaccordingtotheirtemporalresponseinthePSTH.Theacces-sibleparameterforthisprocessingcanbeadjustedfromthispartoftheOptions-tab.Specifytheeigenvectorcalculationusingthepop-up-widgetmarkedwith←ainfigure9.

Figure9:Options-Tab:Classification

Fortheclusteringprocessofthisdatasomeparametercanbeadjustedviathewidgetsmarkedwith←b.Theclassificationprocesscanadditionallyappliedtoallexperimentalfilesinadirectorydefinedin←cinfigure9.Inthiscasetheactualdefinedparametersetinthisandtheothersub-windowsoftheOptions-tab(e.g.thebinsize,numberofselectedelectrodes,etc.)isusedfortheclassificationofthedirectorydata.

SelectionAsetofelectrodesfromtheactualopenedexperimentcanbeselected.Thedatasetrepresentedbythisselectionisthenusedforallfurtheranalysisorvisual-izationmethods.Themainpartofthewindowisoccupiedbya10x10electrodematrix.

Figure10:Options-Tab:Selection

11

Whenanexperimentaldatafileisopenedtheactivityrecordedoneachelectrodeiscolorcoded:darkerbuttonbackgroundsmeanhighactivity;lightergreybackgroundslowerornoactivity.Ifthemousecursorisshortlypausedoveroneoftheelectrodebut-tons,atooltipwiththeactualnumberofrecordedeventsduringthisexperimentonthiselectrodeisshown.IftheAll-checkboxontherightbecomesactivated,allelectrodeswillbeselected.TheInvertSelection-buttoninvertstheactualselection.ActivatingtheCumulativeactivity-checkboxwillenablethecalculationofthecumulativeac-tivityonallelectrodesand,ifappropriate,includeittodataanalysisand/orexportedfiles.

“ElUltimo”ThispartoftheOptions-tabincorporates,asthenamealreadyindicates,accesstofunctionand/orparameterwhichdonotreallyfittoanyoftheotheroption,orwereimplemented“lastminute”totheprogram.Thisdoesnotmeanthatthisfunctionwerenotcomprehensivelytested*well,asfarastimepermitted*-mainlytheprovidedfunctionalityisnotfullyimplemented.ForthefirstreleaseofneurALCtheshufflingfunction(figure11←a)offersjustonewayofdoingit,andmightbesubjecttofurtherextensioninthefuture.Likewisethefunctiontoextractthetimestampsofadesig-natedfiringsequenceofneuronalelectrodes(figure11←b),unitsortemporalresponseclasses-theextractionworks,butdetectionofexistingfirepatternisnotimplementedsofar.Wehavedevelopedasmallandsimpleexternalprogramtodeterminesuchfirepatterns-ifyouinterestedinusingthisprogramandtocollaborate,feelfreetocontactus.

Figure11:Options-Tab:“ElUltimo”

4.2TheneurALCMenuBar

Theprogrammenubar(p.13,figure12)providesaccesstoallimplementedanalysisandvisualizationfunctionsofneurALC.Ifacommandischosenfromanyofthemenus,allrequiredparameterswillbeusedasdefinedinthismomentintheOptions-tab(pp.8).

12

Themenubarisstructuredin5menus:theneuralc-,File-,Analysis-,Contraptions-,andHelp-menu.DependingontheoperatingsystemneurALCisrunningon,themenu,asshowninfigure12,anditsentriesmightlookandbehaveslightlydifferent.Manymenuentriescanbeadditionallynavigatedviakeyboardshortcuts-foralistofavailablekeyboardcommandsforneurALCseeAppendix8.1.

Figure12:Menubar:theneuralc-menu

4.2.1Theneuralc-menu

Thismenu(figure12)providesaccesstotheAboutneuralc-informationandallowstoquittheprogram.4.2.2TheFile-menu

FromthismenuexperimentaldatafilesinNEV2-formatcanbeopenedandanalyzedinneurALC(File→OpenFile).AnyopenedfilescanbesavedtoanewfileinNEV2-format(File→SavetoNEV2).

Figure13:Menubar:theFile-menu

Ifthepost-stimulustimedataoftheopenedexperimentwereclassifiedintoTRCs,

13

theresultcanbesaveusingtheFile→Saveclasses(ASCII)-entryfromthemenu.CalculationresultscanbesavedtofilesinASCII-formatusingtheFile→SaveResults(ASCII)command.IfMySQLwasinstalledonthesysteminuse,andsuccessfullyde-tectedbyneurALC,theresultsoftheactualdisplayedanalysiscanbetransferredandstoredinthedatabaseusingtheFile→Exporttodatabase...-command.ChoosetheFile→Close-commandtoclosetheactualopenedandanalyzedexperimentalrecord-ing.Additionallyadirectorywhichincludesasetofexperimentalrecordingscanbespecifiedandtherecordedresponsescanbeclassifiedoverallexperimentsaccord-ingtotheirresponsecharacteristicsinthepost-stimulustimehistogram(PSTH).TheClassifydirectory...-commandprovidesthisfunctionality.

ViatheImporttrigger-andExporttrigger-commandit’spossibletoimportanex-ternalfilewhichincludes“trigger”informationabouttheappliedstimulation,respectivelytosavethis“trigger”informationofthecurrentlyopenedexperimenttoafile.neurALCpossessesaprogram-internaleditorforthetriggerinformation(p.8)-thiseditorisre-strictedinthesizeofthetriggerinformationitcanhandleanddoesnotofferhighlysophisticatedmanipulationfunctions.Savingthetriggerinformationallowstoextractin-formationfromthisdatabyusingscriptingorprogramminglanguages,orexternaltexteditor.FinallytheFile-menuprovidesprintingfunctionality-anygraphcalculatedanddisplayedbyneurALCcanbeprintedviaFile→Print-command.4.2.3TheAnalysis-menu

Selectinganycommandfromthismenuappliesthechosenanalysistothedatasetvisibleintheselectedtab:ifthePopulation-tabisactive,theanalysisiscarriedoutonpopulationlevel;iftheElectrode-tabisactive,thecalculationiscarriedoutonandvisualizedfortheselectedelectrodeorneuronalunit.

Figure14:Menubar:theAnalysis-menuwiththeOverview-submenuselectedTheOverview&PSTH-submenu(Analysis→Overview/PSTH)Choosinganentryfromthismenu(figure14),appliestheselectedcalculationontheselecteddataandvisual-izestheresultoverthewholeexperimentalduration.Viathissubmenuitispossibleto

14

calculateand/orvisualizethedataasarasterplot(Analysis→Overview→Raster).TheavailableadditionaloptionsAnalysis→Overview→Count,Analysis→Overview→RateandOverview→Probabilityresultinanoverlayinwhichthenumberofspikeperbin(Count),thefiringrateperbininHz(Rate),ortheprobabilityperbin(Probability)iscalculatedandvisualized.IdenticaloptionsareavailablefromtheAnalysis→PSTH-submenuresultinginsimilarformsofvisualizationforthepost-stimulustimehistogramdata.

TheISI-submenu(Analysis→ISI)Thecommandsfromthissubmenuallowtocalcu-lateandvisualizetheinter-spikeintervals(ISI)forthechosendata.Twoadditionalop-tionsareavailable:thebinnedcount(Analysis→ISI→Binnedcount),andthebinnedprobabilityoftheISI(Analysis→ISI→Binnedprobability)canbecalculatedanddis-played.

Figure15:AnalysisMenu:theISI-submenu

TheCorrelation-submenu(Analysis→Correlation)Thissubmenuallowstocalcu-latethelinearcorrelationorautocorrelationfunctionoftheactualselecteddata.

Figure16:AnalysisMenu:theCorrelation-submenu

15

TheInstantfiringrate-submenu(Analysis→Instantfiringrate)Thissubmenuprovidesthefunctionalitytocalculatetheinstantfiringrateonpopulationorelectrodelevel.

Figure17:AnalysisMenu:theInstantfiring-menu

TheSpectrum-submenu(Analysis→Spectrum)Lineartimeseriesmethodwhichcomputesapowerspectrumbybinningadjacentfrequenciesforthebinnedselecteddata.ImplementationusestheFFTW3library.

TheDelay-submenu(Analysis→Delay)Thisfunctioncanbeemployedtounfoldthemultidimensionalstructureoftheselecteddata.Delaycalculationsarecomputedac-cordingtotheoptionsgivenintheOptions-tab.ImplementationtakenfromtheTISEANpackage.

TheMutualinformation-submenu(Analysis→Mutualinformation)EstimatesthetimedelayedmutualinformationofthedataaccordingtotheoptionsgivenintheOptions-tabusingafixedmeshofboxes.ImplementationtakenfromtheTISEANpackage.TheRecurrence-submenu(Analysis→Recurrenceinformation)Thisfunctionpro-ducesarecurrenceplotofthe,possiblymultivariate,dataset.Thatmeans,foreachpointinthedatasetitlooksforallpoints,suchthatthedistancebetweenthesetwopointsissmallerthanagivensizeinagivenembeddingspace.ImplementationtakenfromtheTISEANpackage.4.2.4TheContraptions-menu

Thismenu(figure18)providesaccesstoavarietyoffunctionswithdifferentpurpose:usetheAnalysis→Contraptions→Saveconfiguration-commandtosavetheactualparameterdefinedintheOptions-tabtotheconfigurationfileneuralc.conf.ThisfileislocatedinsidethefolderinwhichtheneurALCwasinstalledandreadatprogram

16

startup.TheContraptions→Select3Dcolormap-commandallowstoloadanduseauser-definedcolormapforthevisualizationsinthe3D-tab4.1.3.

Figure18:Menubar:theContraptions-menu

TheContraptions→Connecttodatabase-commandallowstoconnecttoaMySQL-servertotransferandstoreexperimentaldatafilesandresults.Selectingthecommandopensawindowinwhichtherequiredinformationfortheconnectioncanbedetermined.

Figure19:Contraptionsmenu:theConnecttodatabase-dialogue

Usethethetextentryfieldsshowninfigure19todefinethenameoftheexistingdatabase,thehostnameonwhichtheMySQL-serverisrunning,theusernameandpassword,andtheportonwhichthedatabaseserverallowsconnections.IfneurALCsuccessfullyconnectstoaMySQLdatabase,theContraptions→OpenDB-GUI-commandcanbeusedtoopentheprovidedgraphicaldatabaseuserinterface(p.26).4.2.5TheHelp-menu

ThehelpmenuinthemenubarcanbeusedtoopenneurALC’sonlinehelp.

17

5Analysis&VisualizationBackground

5.1neurALC’sAnalysisWorkflow

Thissectiongivesashortoverviewonthebackgroundofthedifferentanalysismeth-odsusedinneurALC.Theprogramincorporatesmethodstakenfromlinearlineartimeseriesanalysis(spectrum,auto-andcross-correlation,histograms,etc.)andlow-dimensionalchaosconcepts(recurrence,delay,etc.),whichhasproventobefruitfulintheunderstandingofmanycomplexphenomena(despitethefactthatveryfewnatu-ralsystemshaveactuallybeenfoundtobelowdimensionaldeterministicinthesenseofthetheory).Itisofcourseuptothescientistwhodoestheanalysistoputtheseresultsintotheirpropercontextandtoinferwhatinformationsheorhemayfinduse-fulandplausible.Figure20illustratesneurALC’sunderlyingworkflowforanalysisandvisualization.

Figure20:neurALC’ssimplifiedanalysisworkflow

Theuserselectsasingleormultipledatafiles,predeterminesallrequiredparametervianeurALC’sOptions-tab(p.8),andselectsthedesiredanalysisandvisualization.Thespecifiedcalculationiscarriedoutandtheresultcanbedisplayed,storedinanASCII-file,ortransferredandstoredinaMySQLdatabase.Thedifferentcoloredarrowsrepresentsthetwo,basicworkflowsinneurALC:an“interactive”pathbasedontheGUI,

18

whichallowsexplorationofaloadeddatafile,connectedbythebluearrows.Itismainlyintendedtoadjustandfindtheanalysisparameterasneeded,butallowsaswelltosaveindividualvisualizations,ortheresultsforaselectedensembleofelectrode-,neuronalunit-,ortemporalresponseclass-data.

changesIfnotexplicitlysavedusingtheContraptions→Exportconfiguration-commandactualneurtoanyALCparametersession.

intheOptions-tab(pp.8)areonlyvalidandhenceusedduring(p.16),theConnectedbytheblackarrowsthe“semi-automatic”pathisshown,whichusesthepreviously,interactivelydeterminedparameterandrunswithminimaluserinteractions,executingtheselectedcalculationforasetofdatafiles,electrodes-orneuronalunits-selection,andsavesthecomputedresultstoexternalfile(s)forfurtheranalysis.Thespecifiedcalculationsorvisualizationcarriedoutinthis“path”willusetherequiredparameterasdefinedwithintheOptions-tab(p.8)inthemomentofanalysis-thisim-portantconceptualaspectiselucidatedbythethick,blackstrokearoundtheOptions-andAnalysis-boxinfigure20.

electrode-,Allavailablecalculationsandvisualizationscanbecarriedoutaswellonpopulation-,asonarethereforeneuronalGUIcontext-sensitive:unit-,ortemporaltheresponsevisualizationclasslevel.dependsSelectedonthecalculationsactualselectedorvisualizationstabintheneurALCGUI..

5.2SpikeTrainRepresentation

neurALCoffersafewwaysofrepresentingtherecordeddata.Basicallydatacanbedisplayedonpopulation-(p.5)orindividualelectrode-,neuronalunit-,ortemporalre-sponseclass-level(p.6).Inthiscontextspikedatasetcanbedisplayedoverthefulltimecourseoftheopenedexperiment(Analysis→Overview,p.14),ordisplayedaspost-stimulustimehistogram(Analysis←PSTH,p.15).Rasterplots&ActivityEstimation

OverviewAsdefaultexperimentaldataisdisplayedinformsofarasterplot(figure21).Eachrecordedspikeeventisdisplayedusingasingleblackdot.Alongtheabscissatheexperimentaltimecourseisrepresented,alongtheordinatetheactiveelectrodes(labeledasChannel),whichloggedtheneuronalactivity,areshown.Usingtheleftmousebuttonitispossibleinany2D-displaytozoominsidethedataplot-figure21,forexample,showstherecordedactivityon100activeelectrodesforthetimeintervalbetweenthe15thtill30thsecondoftheexperimentusingthisfunctionality.Forthisexampleaperiodicstimulationwithaperiodlengthofapprox.2swasused,resultinginthecorrelatedneuronalfiringpatternshowninthefigure.

Ifmouse-basedaSelectingparticulartheregionFile→ofPrintthedata-commandrepresentationallowstowasprintmagnifiedtheactive(asdatashowndisplayininsideneurALC.supportedbyZoom-function,theOS,savedtoonlyafile.)thisarea.

oftheplotwillbeprinted(or,iffigurethisfunctionality21)usingtheis19

Raster

1008060402001618202224262830Time (s)

Figure21:populationrasterplot

Post-StimulusTimeHistogram(PSTH)Anotherwayofcharacterizingtherecordedneu-

ronalresponsepatternisprovidedinformsofthepost-stimulustimehistogram.Usingthetimetofthestimulus,thespikesaredisplayedinawindow∆τalignedatt.

PSTH -- Count

30252015105000.511.52Time (s)

Figure22:populationPSTHplot

Thewholeexperimentalrecordingshownpartlyinfigure21isplottedinthiswayinfigure22.Theblackdotsrepresenteachrecordedevent,alongtheabscissathetimeinterval∆τalignedtotheoccurrenceofthestimulusatt,markedbytheredsmallcrosses,isshown.Alongtheordinatetheeachpresentationofthestimulationisrepresented.Dependingontheinforma-tionprovidedwithintheexperimentaldatafile,∆τcaneasilyadjustedusingthefunctionalityprovidedintheTrigger-options(p.8).

20

ActivityEstimationOverlaysneurALCprovides3methodstoestimatethepopulation-or

electrode/neuronalunit-activityfromthedatashownintherasterplots.Byapplyingabinningprocedureitispossibletocalculatethesimplecount(spikes/bin),rate(inHz),orprobability/binoftherecordedneuronalevents(see.pp.14).Theresultoftheselectedcalculationisthendisplayedasanoverlayintheactualrasterplot,exemplaryshowninfigure22forthespikecount(superimposedtotherasterwithablueline;binsize=1ms).

ISIThisfunction(p.15)allowstocalculatedanddisplaytheinterspikeinterval(ISI)distribution

ISI -- Count76543210500100015002000forthepopulationorthePSTHdataoftheselectedelectrode-,orneuronalunit-data.Only

Interval (ms)

Figure23:exampleofaPSTH-basedISIdistributionofasingleelectrode

interspikeintervalsoftheactualdatasetwhichmatchthelimitsdefinedintherelatedOptions-tab(p.9)areextractedandusedtocalculatetheintervaldistribution.InstantFiringRateCalculatestheinstantspikefiringrater(t)forthepopulation-,orse-lectedelectrode-orneuronalunit-data(p.16).Theoriginalspiketimedataisdividedintobins,

Instant Firing Rate PSTH1412108642000.511.52Time (s)

Figure24:instantfiringratefunctionforthePSTHdataofasingleelectrode

eachofwidth∆τ.Thusifonelooksonabincenteredontimet,afunctionn(t)canbedefinedwithN=1ifthereisaspikeinthebinandn=0ifthereisnot.Whetheraparticularspiketi

21

isinabinwithsize∆tattimetcantosuchanextentevaluate.Tofindanyspikeinthedefinedbinsallpossiblespikestobesummed,sothatthefunctionn(t)thatcountsthespikescan󰀁havet−tibewrittenasn(t)=f[∆τ].FromthesumineachbinthenthefiringrateinHzineachiscalculatedandplotted.

i

SpectrumComputesapowerspectrumofthebinneddatabybinningadjacentfrequencies.

ThisroutineusestheFFTW3libraryandbydefaultzeropadding,thatis,noperiodiccontinua-tionisassumed.

5.3MultivariateTimeSeriesAnalysisMethods

5.3.1Correlation

Computestheauto/crosscorrelationfunctionofabinnedspiketimeseries(p.15).

Crosscorrelation0.20.10-0.1-0.4-0.200.20.4Time (s)

Figure25:exampleofacrosscorrelationfunctionofasingleelectrode

Thisfunctioncomputesthecrosscorrelationsbetweentwobinnedtimeseries.AutoorcrosscorrelationfunctionsarecalculatedsymmetricallyforthethewindowlengthdefinedintheOptions-tab(p.15).Thedefaultvalueis0.5s,sothatthecrosscorrelationfunctionshowninfigure25iscalculatedandvisualizedwithinthetimewindow[-0.5:0.5]s.

5.3.2Recurrence

Recurrenceplotsareausefultooltoidentifystructureinadatasetinatimeresolvedwayqual-itatively.Thiscanbeintermittency(whichisdetectablealsobydirectinspection),thetemporaryvicinityofachaotictrajectorytoanunstableperiodicorbit,ornon-stationarity.Theywerein-troducedin[Eckmann1987,Casdagli,1997],whereyoufindmanyhintsonhowtointerprettheresults.Foreachpointinthedatasetitlooksforallpoints,suchthatthedistancebe-tweenthesetwopointsissmallerthanagivensizeinagivenembeddingspace(p.10).Thetimeseries,e.g.theestimatedpopulationactivityasbinnedcount,issimplyscannedandeachpairofthetimeindices(i,j)whosecorrespondingpairofdelayvectorshasadistance≤󰀒ismarkedwithadot.Thusinthe(i,j)-plane,dotsindicatecloseness.Inanergodicsituation,thedotsshouldcovertheplaneuniformlyonaverage,whereasnon-stationarityexpressesitselfbyoveralltendencyofthedotstobeclosetothediagonal.Ofcourse,areturntoadynamicalsituationthesystemwasinbeforebecomesevidentbyahighdottedregionfarawayfromthe

22

diagonal.Toachieveanagreeabledisplayspeedfortheresultofthisanalysis,thenumberofpointscalculatedisdecimatedfortheon-screenvisualization.TheimplementationinneurALCistakenfromTISEAN.

5.3.3Delay

ThetimeevolutionofspikingneuronalsysteminsomephasespaceΓ⊂Rdcanbeex-pressedindiscretetimet=n∆tbymapsoftheformxn+1=f(Xn).Neuronalrecordingscanthenbethoughtofasasequenceofobservations{sn=s(Xn)}performedwithsomemeasurementfunctions(·).Sincethisscalarsequence{sn}initselfdoesnotproperlyrep-resentthemultidimensionalphasespaceofthedynamicalsystem,thisfunctioncanbeem-ployedtounfoldthemultidimensionalstructureoftheavailabledata.Themethodofdelaysrepresentsoneofthemostimportanttechniqueforphasespacereconstruction.Vectorsintheembeddingspace,areformedfromthetimedelayvaluesofthescalarmeasurementssn=(sn−(m−1)τ,sn−(m−2)τ,...,sn).Thenumbermofelementsiscalledtheembeddingdi-mension,thetimeτisgenerallyreferredtoasthedelayorlag.TheimplementationinneurALCistakendirectlyfromTISEAN.

Delay403020100010203040-

Figure26:delayrepresentationofapopulationrecording(delay=25ms,dimensions=2)

5.3.4MutualInformation

Thisfunctionestimatesthetimedelayedmutualinformation[FraserandSwinney,1986]ofthebinnedselecteddatausingafixedmeshofboxes.Unliketheautocorrelationfunction,themutualinformationtakesinaccountalsononlinearcorrelations,andisintendedtodeterminea󰀁pij(τ)

whereforsomepartitionontherealreasonabledelay(p.10).ComputingS=pij(τ)lnpipj

ij

numberspiistheprobabilitytofindatimeseriesvalueinthei-thinterval,andpij(τ)isthejointprobabilitythatanobservationfallsintothei-thintervalandtheobservationtimeτlaterfallsintothej-th.Forembeddingdimension≤2existgoodargumentsthatifthetimedelayedmutualinformationexhibitsamarkedminimumatacertainvalueofτ,thenthisagoodcandidateforareasonabletimedelay.TheimplementationinneurALCistakenfromTISEAN.

23

5.43-DVisualization

neurALCofferssomevisualizationfunctionsinthree-dimensionalspace.Selectthe3D-tab(p.7)andchoosePSTHfromtheOption-tabtovisualizethecumulativebinnedPSTHoftheelectrodesetdefinedintheglobalOptions-tab(p.11)in3D(figure27).Asinthe2D-display(p.15)thePSTHintervalisrepresentedalongtheabscissa;theordinateshowsthenumberofstimulusrepresentationschronologicallyorderedfromtoptobottom.Thebinnedspikenumberisdisplayedusingauser-definablecolormap(p.7).Usingthemousethedataviewcanberotated,zoomed,orpanned(seefulllistofkeyboardshortcuts&mousenavigationoptionspp.35).

Figure27:defaultviewofabinnedcumulativePSTHvisualizedin3D

Additionallytwoformsofaspectrogramcanbecalculatedandplotted:forthefullexperi-mentalcourse,orforthePSTHdataoftheactualselection(usetheOption-pop-upmenuinthe3D-tab(p.7)toselectthedesiredvisualization).Tocalculatedthespectrogram,thebinned

Figure28:defaultviewofaspectrogramcalculatedforthePSTHofaselectedelectrode

neuronalresponseisdividedinwindowswithapresetlengthandoverlap(p.9).Nextafouriertransform(FFT)isappliedtothedataineachwindow,derivingthefrequenciesandamplitudesofitscomponentsimplewaves.DependingonthesizeofthefourieranalysiswindowusedfortheFFTanalysis,differentlevelsofresolutioncanbeachieved.Alongwindowresolvesfrequen-ciesattheexpenseoftime-theresultisanarrowbandspectrogram,whichrevealsindividualcomponentfrequencies.Ifasmallanalysiswindowisused,adjacentfrequencycomponentsaresmearedtogether,buttheresultdisplaysabettertimeresolution.Spectrogramsasshown

24

inwindowfigurethewhich28,representisusedthetoderivethetimethealongfrequencytheabscissa.components,The“time”whichunitareusedthenisrepresentedthenumberofalongthemomentordinatethisaxis.visualizationTheamplitudestechniqueofistheintendedfrequenciesonlyasareaqualitativelydisplayedcolormeasure.

codedinz.Forthe5.5Classificationofelectrode&“unit”responses

5.5.1ConceptofTemporalResponseTypeClasses(TRC)

neurtimeALC(TRC).courseoffersintheamethodPSTH.toTheclassifyresultingtherecordedclassesareneuronaldenominatedresponsepatternaccordingtotheirthemanagepost-stimulusSuchaclasstimedescribeshistogram.theSimilarthebinnedbinnedtimeresponsescourseareoftheasassignedresponsetemporalresponseclasstotheassamecalculatedTRC.Toinfollowinglargesequentialdatasetsapproachofneuronaltoclassifyrecordingstheresponsesandtorepresentintotemporalthisdata,responsetheprogramclasses:usesthe1.ReducingcipalComponentthedimensionalityAnalysis(PCAofthe).binnedpost-stimulustimehistogramdatausingPrin-2.Clusterfromtheanalysisfirststep(basedbyplacingonKlustaKwik)themintonewlywhichassignedreducestheTRCsnumber.oftheobjectsresultingofIfwastheyou’rerecordedsearchingspikeforsignals,aprogramyoumightwhichtakedoesalook“classical”onanother“unit”-sortingfreebasedonthewaveformsignalsdevelopedinourlab.NEV2lkitisaspikepreprocessor,whichallowsprogram,theextractionNEV2lkit,whichtheirASCII-based,signalfromshapecontinuousrecordeddataand/orneuronalspikeclassificationinaccordanceofspikewithore.g.(waveNEV1.xform).filesNEV2lkittoNEV2.0isforadditionalfurtherrecommended,usagewithneurifALC.

youwanttoconvertPrincipalcomponentAnalysis(PCA)TheestimatedactivitysignalinthePSTHrep-resentalargenumberofvariableswhichmakesitverylikelythatsubsetsofthesevari-ablearehighlycorrelatedtoanother.Theaccuracyandreliabilityofaclassificationofsuchdatawillsufferifhighlycorrelatedorvariableswhichareunrelatedtotheoutcomeareincludedtotheanalysis.

ReducingthedimensionalityofthedatainthePSTHwithoutsacrificingaccuracyisthereforeakeystepinthisclassification.PCAtransformsanumberof(possibly)correlatedvariablesintoasmallernumberofuncorrelatedvariables-theprincipalcom-ponents.Asaresultthedimensionalityofthedatasetisreducedbutmostoftheoriginalvariabilityretained.Thefirstprincipalcomponentaccountsforasmuchofthevariabilityinthedataaspossible,andeachofthesucceedingcomponentsaccountsforasmuchoftheremainingvariabilityaspossible.

neurALCperformsPrincipalComponentAnalysisonthePSTHdataconstructedforeachselectedfileanditscontainedelectrodeorneuronal“unit”data.Itcalculatesthefirst3principalcomponentswhicharethenusedinsubsequentclusteranalysisbasedonKlustaKwik.Usingtherelatedpop-upmenufromtheClassification-windowintheglobalOptions-tabintheGUI(p.11),itispossiblytospecifywhethertheeigenvectorcalculationisbasedonthecorrelations,thevariance/covariancesorthesumofsquaresofthecross-productsofthebinnedPSTHsignalmatrix.

25

ClusteringLaststepoftheunitsortingistheclusteringofthedata.neurALCusesKlustaKwik,aprogramforunsupervisedclassificationofmultidimensionalcontinuousdata.KlustaKwikdeliversamongothers:fitamixtureofGaussianswithunconstrainedcovariancematrices,choosesautomaticallythenumberofmixturecomponents,andrunsfastonlargedatasets.KlustaKwikisbasedontheclassificationexpectationmaximizationalgorithm(CEM)[CeleuxandGovaert,1992].DefinableparametersfortheclusteringprocessintheneurALCGUIare:

MINCLUSTERStherandominitialassignmentwillhavenolessthannclusters.The

finalnumbermaybedifferent,sinceclusterscanbesplitordeletedduringthecourseofthealgorithm.Thedefaultvalueis1.MAXCLUSTERSdefinesthemaximumofpossibleclustersn.Clustersplittingcanpro-ducenomorethannclusters.Thedefaultvalueis5.PENALTYMIXitispossibletospecifyBayesianinformationcontent(BIC)orAkaikein-formationcontent(AIC)asthepenaltyforalargernumberofclustersoramixtureofthesetwo.ThiswidgetallowstodefinetheamountofBICtouseapenaltyformoreclusters.Defaultof0setstouseallAIC.Use10touseallBIC(thisgenerallyproducesfewerclusters).AllotherparameterswhichareknowninthestandaloneversionofKlustaKwikprogramcanbeeditedinthesourcecodeofneurALC.

6MySQLDatabaseIntegration

neurALCprovidessomedatabasefunctionalityusingMySQL.MySQLisanopensourceSQLdatabasewhich,quotingtheMySQLwebsite,is“designedforspeed,powerandprecisioninmissioncritical,heavyloaduse.”WithintheMySQLdatabaseoriginalex-perimentfiles,analysisresultsandtheinformationoftheappendantparameterused,modifieddata,comments,orevendigitalphotographsoftheneuronalspecimen,whichwasusedintheexperiments,canbestoredandadministered.neurALCintegratesseamlesslywiththeMySQLclient-resultsofcalculations,comments,originalexperi-mentaldatafilesandanymodification,etc.aretrackedandautomaticallytransferredtothedatabase.Asecond,supplementaryprogram-dbALC-isprovided.Itoffersasimplegraphicaluserinterfacetoaccess,administerandretrievedatastoredintheMySQLdatabase.

6.1TechnicalPrerequisites

TouseneurALC’sdatabasefunctionality,aninstallationofMySQLversion3orhigherisneeded.Pleasevisithttp://www.mysql.comanddownloadthepackageofyourchoice.neurALC1.0.0wassuccessfullytestedwithMySQL3,4and5.

AdditionallytheQt3librariesonyoursystemmustofferMySQL-driversupport(thesup-pliedbinaryinstallersofneurALCforApple’sMacOSXandMicrosoftWindowsinclude

26

theQt3libscompiledwiththerequiredsupport.UnderLinuxanupdateviathepackagemanagerofyourdistribution,orevenarecompilationoftheQt3C++frameworkmightbeneeded).

InstallMySQLonyourcomputersystem-fromourexperience,dependingontheoperatingsystemandthepackageofMySQLyouchoose,thismightbethemostprob-lematicpart.BeforeyoutrytousethedatabasefunctionalityprovidedbyneurALCordbALC,makesurethatyoucheckthecorrectfunctionoftheMySQLinstallationonyoursystem.ReferfortheMySQLinformationwhichcomeswiththepackageyoudownloaded,ortheonlinedocumentationavailableathttp://www.mysql.com/documentation.

IfMySQLissuccessfullyinstalledandrunning,addauser“neuralc”withthepass-word“neuralc”tothedatabase(refertotheMySQLdocumentationforinformationaboutaddingauserandpassword).ThisisthedatabaseaccountwhichisusedasdefaultfromwithinneurALC.Inthenextstepadatabasewiththerequireddesignandtabletype“InnoDB”mustbecreated.Theeasiestoptionistocopythefilesshowninfigure29(providedwiththeneurALCdistributioninsidethefolderneuralcDB)intoMySQL’sdefaultfolderfordatabases.ThenexttimeyoustartupMySQLthesefilesarefound

Figure29:ContentofthefolderneuralcDB

andcanbeaccessed.AlternativelyyoucanopenthefileDBcreation.txt(figure29,labeledingreen)intheeditorortextprocessorofyourchoice,andusethedocumentedcommandsequencefromwithinMySQLtocreatethecorrecttablesandkeys(there-quiredcommandlogisaswellincludedinsection6.4,p.30).

6.2DatabaseDesign

ThedatabasedesignwhichisusedbyneurALCissketchedinfigure30.Whileitissurelynotthemostsophisticateddatabasedesign,it’ll*hopefully*suitsthe“minimal”requirements.

Twomaintables,experimentandcalculusareemployed.experimentisrelatedwiththefieldsinwhichtheoriginalexperimentaldatafileinNEV2-format(nev_file),acom-ment(comment),thecreationdata(creation_date),adigitalphotograph(photo),the

27

experimentrelatedstimulusprograme.g.aPythonscript(stimulus_program),orthefilename(file_name)isstored.Inshortthesetablesholdallexperimentrelatedinfor-mationapartfromtheinformationwhichisrelatedtoanyappliedanalysis.

Analysisresultsbecomerepresentedinthecalculusentitityandrelatedtables(markedinthefigurebyadarkgreybackground),whichholdthisanalysisrelatedinformation.

Figure30:ExtendedEntityRelationDesignofthedatabaseusedbyneurALCDifferenttablesforstoringtheresultsandrelatedanalysisparameterarecreated:theelectrodenumber(elec_id),ifavailabletheunitinformation(unit),thestatusoftheTRCclassification(classified),recurrencerelatedinformation(boxesandvecinity),theresultofthedelaycalculation,theembeddingdimensionused,theappliedbinstepwidth(bin_step),thefrequencyanalysisrelatedinformationabouttheusedwindowsizeandpossiblewindowoverlapinthiscalculation,theISI-relatedinformation(minor_ISI,major_ISItion(data_id),the).Additionallyresultofanone,appliedsofarbinningunused(bin_datakey(unused),and)isthecreated.

generaldatainforma-6.3dbALC-adatabaseGUI

WhileexperimentaldatafilesandrelatedresultscanbetransferredtotheMySQLdatabase,anadditionalprogramdbALCisintendedformoreconvenientdataretrieval.Itprovidesasimplifiedgraphicaluserinterface(figure31,p.29)tobrowsethedata,andretrievethestoreddata.

ThedbALCcanbedividedinfivefunctionalareas,labeleda-einfigure31.Alistview

28

Figure31:ThedbALCGUI

displaysallactuallysavedexperimentswithintheneurALCdatabase.Anyexperimentscanbeselectedfromthiswindow.Thepop-upmenulabeledwithCalculus(d)allowstoselectthespecifictypeofanalysisresultwhichisavailablefromthisexperiment.Fur-thermoreanyselectionfromthispop-upwilldynamicallychangethelistofexperimentsdisplayedina.Thismeansthat,ifforexamplePSTHisselectedfromthepop-up,thelistofexperimentswhichcanbeselected,becomesrestrictedtotheoneswhichincludethistypeofanalysisresult.

Thetextentryfieldsinbdisplaytherelatedparameterinformation,orallowtoenterdirectlyaspecificparameter,forwhichthentheexperiments,whichcontainthisinfor-mation,willbefilteredfromthedatabaseanddisplayedinthelistview(a).

Analysisresultswillbedisplayedinthetableviewlabeledwithc.Inthecolumn-viewthestoredresultsforeach,forthedefinedanalysischosenelectrodeorneuronalunit(seep.11)willbedisplayed.Theselectedanalysisresultscanbeextracted(ordeleted)usingtheExtract...-buttonine.ActivatingtheExtract.nevfile-checkboxwilladdition-allyextractthecorrespondingexperimentaldatafileinNEV-formatfromthedatabase.

29

6.4SQLcommandlogforcreatingtheneurALCdatabase

CREATEDATABASEneuralcDB;GRANTSELECT,INSERT,UPDATE,DELETEONneuralcDB.*

TO’neuralc’@’localhost’IDENTIFIEDBY’neuralc’;FLUSHPRIVILEGES;

CREATETABLEexperiment(file_nameVARCHAR(256),nev_fileLONGBLOB,commentVARCHAR(256),creation_dateDATE,photoLONGBLOB,

stimulus_programLONGBLOB,PRIMARYKEY(file_name(256)));

CREATETABLEcalculus(data_idENUM(’None’,

’PopulRaster’,’PopulCount’,’PopulRate’,’PopulProb’,’PSTHRaster’,’PSTHCount’,’PSTHRate’,’PSTHProb’,’ISICount’,’ISIBinCount’,’ISIBinProb’,’IFRPopulation’,’IFRPsth’,’Autocorr’,’Crosscorr’,’Spectrum’,’Delay’,’Mutual’,’Recur’,’PSTH3D’,

’SpecPopul3D’,’SpecPSTH3D’),

elect_idSMALLINT(3),unitSMALLINT(3),

classifiedTINYINT(1)UNSIGNED,boxesSMALLINT(4)UNSIGNED,vecinitySMALLINT(4)UNSIGNED,bin_stepSMALLINT(4)UNSIGNED,delaySMALLINT(4)UNSIGNED,windowFLOATUNSIGNED,

overlapSMALLINT(4)UNSIGNED,dimmensionSMALLINT(4)UNSIGNED,minor_isiSMALLINT(5)UNSIGNED,major_isiSMALLINT(5)UNSIGNED,bin_dataLONGBLOBNOTNULL,unusedSMALLINT(5),

experimentVARCHAR(256),

FOREIGNKEY(experiment(256))REFERENCESexperimentONDELETECASCADE,

PRIMARYKEY(experiment(256),data_id,elect_id,unit,classified,boxes,vecinity,bin_step,delay,window,overlap,dimmension,minor_isi,major_isi,unused));

30

7ARealWorldExample

Thissectionprovidesashort“realworld”example.Whenopeninganmulti-electroderecordinginNEV2-format,neurALCdisplaysasdefaulttheopeneddatainthePopulation-tab,offeringanoverviewoftheexperiment.Additionallythepopulationactivityises-timated(usingtheactual,intheOption-tabdefined,binsize)anddisplayedasabluelinedCount-overlay,representingthenumberofspikescountedineachbin.

Figure32:RasterplotoffullexperimentaltimecourseandpopulationactivityoverlayInthelowerpartofthewindow,themaximumofthebincountobservedintheopenedexperimentaldatafileisdisplayed(Maximum:55infigure32).Thesliderwidgetatthebuttoncanbeusedtointeractivelyadjustthebinsize-thepopulationactivityoverlaywillbeimmediatelyrecalculatedandthedisplayupdated.Usingthemenu-barprovidedanalysisfunction(p.14)thepopulationactivitycanalsobedisplayedintermsofitsfrequencyorprobability.

Figure33:Zoomedpopulationactivity

Usingthemouseitispossibletozoomthedisplay(seesection8.2).Fortheexample

31

shownhere,thezoomeddisplayallowseasyvisualinspectionoftheperiodicmodula-tionofthepopulationactivity.Clickingwiththeleftmouse-buttoninsidethedisplayareasetsalabelwiththetimeandchannelcoordinate.

Choosingtherecurrence-function(p.22)arecurrenceplotfortherecordedneuronalactivityinthisexperimentisdisplayed(figure33).Lightershadedregionsinthisplotindicatethatthesedataaredynamicallydistinctfromtherest(seep.22formoreinfor-mation).Inthelowerpartofthewindowthebinsizeusedforthecalculationisdisplayed(binsize:50ms,takenasdefinedintherelatedwindowintheOptions-tab).Thesliderwidgetatthebottomcanbeusedtointeractivelyvarythevicinity-parameter.

Figure34:Zoomedrecurrenceplotforthebinnedpopulationactivity.

Theunderlyingperiodicitycanberevealedbycalculatingtheautocorrelationfunction(p.22)forthebinnedpopulationactivity.Fromthevisualinspectioninfigure32and33thewindowlengthfortheautocorrelationcalculationwasadjustedto2.25sintherelatedwindowoftheOptions-tab(p.10).

Figure35:Autocorrelationfunctionofthebinnedpopulationactivity.

Theautocorrelationfunctionshowsamaximumatapprox.2s(peakamplitudeis

32

locatedat1.998s,identifiedbysimplyclickingthetipofthepeakattherightinfigure35).WenowswitchtotheElectrodes-tab(p.6)andselecttheRate-entryfromthePSTH-commandintheAnalysis-menu(pp.14).Figure36.

Figure36:Post-stimulustimehistogramforElectrode1withRate-overlay

FromtheupperleftChannel-pop-upthefirstelectrodeinthisfilewaschosen.Thispop-updisplaystheactualelectrodeselection,aswellasthetotalnumberofspikesrecordedduringtheactualexperimentonthiselectrode-inthisexample:151spikesonelectrode1-1(151).

Thespikesoftheexperimentaldatashowninthisexamplewerepreviouslysortedinto“units”accordingtotheirsignalformusingNEV2lkit,aspikedatapreprocessordevel-opedinourlaboratory.Ifsuch“unit”sortingwasapplied,theUnit-listviewintheupperrightdisplaysalistofthesorted“units”onthiselectrode.Inthescreenshotinfigure36unit3wasselected(markedbythelightbluebackgroundinthelist).Righttothelistviewthenumberofspikes,95,associatedwiththisunitisshown.Belowthisinfor-mationatextfieldinformsaboutthestatusofanadditionalclassificationofthetemporalresponsesoftheseunits(here“Unclassified”;seepp.25).

The2D-plotareaofthewindowshowsthePSTHoftherecordedspikesonthiselec-trodeasarasterplot.Thesmallredcrossesinthedisplayareamarkthetriggerin-formation(seep.8)ofthestimulationappliedduringthisexperiment.ThegreenlineoverlayshowstheestimatedelectrodeactivityforthePSTHdata.Thepinklineoverlay,whichpartlymasksthegreenline,representstheactivityoftheselected“unit”.Thesliderwidgetatthebottomcanbeusedtochangethebinsize,andwithitinteractively

33

theestimatedelectrode,andselected“unit”activity.UsingtheISI-commandfromtheAnalysis-menu(p.21)theinter-spikeintervaldistributionfortheselectedcanbevisu-alized.

Figure37:ISIdistributionfortheselectedneuronal“unit”

Inthenextstepthecrosscorrelationofthisneuronal“unit”withthebinnedpopulationactivityisvisualizedinthenextfigure(binsize=1).

Figure38:CrosscorrelationfunctionforElectrode1,Unit3withpopulationactivityInthelowerpartofthewindowthemaximumifthecalculatedcorrelationfunctionisshown(thisvalueishighlydependingontheactualbinsizeusedforthecalculation!).ToexporttheanalysisresulttoanASCII-fileforfurtherusage,usetheSaveResults(ASCII)-commandfromneurALC’sFile-menu(p.13).IfyouselectedpreviouslyagroupofelectrodesintheSelection-windowoftheOptions-tab(p.11)theselectedanalysiswas(orwillbe)carriedoutfortherelateddataset.Accordinglytheresultsof

34

theanalysisforalltheselectedelectrodesand,ifexisting,neuronal“units”ortemporalresponseclasseswillbeexportedandsavedintheASCII-file.

8Appendix

8.1KeyboardShortCuts

TheQtlibrarymapsautomaticallythemodifierkeytothedefaultoftheoperatingsystemneurALCisrunningon.Asaresultthecommand-keyunderMacOSX(usedforillustrationintheabovefigure)isreassignedtotheControl-key(Ctrl)underLinux,ortheAlternate-key(Alt)underMicrosoftWindows.

35

8.2MouseFunctionality

HoldingdowntheShift-buttonwhilerotatingthe3D-viewrestricts,dependingonthemousemovement,therotationtothey-axis.Ifyouuseaone-buttonmousethe3D-viewpanningispossiblebydraggingtheviewwhileholdingdowntheOption-key.

8.3SupportedFileFormats

8.3.1Input

neurALCversion1.0.0canonlyreadneuronalmulti-electroderecordingsinNEV-fileswhichcomplywiththeNEVspecification2.0.Theprogramcheckstheintegrityofthedataintimespacewhenanexperimentaldatafileinthisformatisopened.Dependingonthedataacquisitionitispossiblethatspiketimesarenotfiledinascendingtemporalorder.neurALCvalidatesand,ifneeded,alignsthisorderwhenopeningafile.8.3.2Output

neurALCversion1.0.0supportsafewoutputformats:

•AllopenneuronaldatafilescanbesavedinNEV2.xformat,whichallowstosaveTRCclassificationandmodifiedcommentstothesefiles.

•Analysisresultscalculatedfortheselectedelectrodeorneuronal“unit”data,canbeexportedandsavedtoanASCII-file.FileswritteninASCII-formatbyneurALChavethefollowing,generalstructure:aheaderofseverallines,followedbycolumnswhichholdtheanalysisresults(figure39).

36

Figure39:ExampleASCII-exportforaselectedgroupofneuronal“units”

Eachheaderlinebeginswitha#.Thisheaderholdsinformationabouttheexperimen-taldatafileusedfortheanalysisresultsandtherelatedcalculationparameter.Thelastlineoftheheaderholdsthe“variables”orlabelsorthefollowing,tabulator-separatedresultcolumns.

9Bugs

Bugs?YoureallyaskingforknownBUGS?!??Geesh...well,wetriedtomaketheprogramasbug-free,aspossible.Adescriptionoftheknown“oddities”ofneurALCisprovidedwithintheneurALC-distributioninthefilebugs.txt.Ifyoudiscoverabug,pleasereferforreportingtothefollowingFAQ.

10FrequentlyaskedQuestions

1.“I’musingtheZimbutsu3000-XXL(autumnof1984edition)data-acquisitionsys-tem(orwhatever)whichsavesdatainitsownfileformat.neurALCinthecurrentversionseemsnotcapabletoopenthesefiles.Whentherewillbesupportforthisfileformat?”-Well,thisisGPL’edsources-goaheadandextendtheprogramforyourself.Ofcourseweareopenforsuggestions,recommendations,etc.-feelfreetoemail(butpleasedonotexpectimmediate*well,ifever*implementation.)2.“Ihavefoundabug.WhatshouldIdo?”-Youcanreportallbugs(orpossiblefixes)throughhttp://neuralc.sourceforge.net.Pleaseincludetothedescriptionthenameoftheoperatingsystemyouareusing,aswellastheQt-,QWT-,QWT3D-,andFFTW-versions.3.“IhavesuccessfullycompiledneurALCandbuildaninstallerontheSony’sPSP,Qtopia,whateverelsesystemand/orplatformforwhichsofarnobinaryinstalla-

37

tionisavailable.DoyouliketoincludeitintheneurALCdistributionstree?”-YES!Contactusbyemailorthroughhttp://neuralc.sourceforge.net

4.“Ihaveextendedtheimport/outputand/oranalysiscapabilitiesofneurALC.DoyouliketoincludethechangestotheneurALCsourcetree?”-YES!Contactusbyemailorthroughhttp://neuralc.sourceforge.net5.“Well,Trolltechjustnowreleasedthemarvelous,wowed,effulgent,*pleaseinsertyourdescriptiveexpressionhere*Qt4-includingevenafreedesktopeditionforMicrosoftWindows.Howcanyoudaretoofferafreeprogrambasedonthean-cient,hoary,paleolithic,*pleaseinsertyourdescriptiveexpressionhere*Qt3?”-WewereandareawareofthereleaseofQt4(andtheresultingreliefthishasforexampleonthedevelopmentofQt-basedprogramsunderMicrosoftWindows).neurALCispartlybasedonextensionstotheQt3framework,namelytheQTWandQTW3Dlibraries.WhenwestartedthedevelopmentneitheroftheselibrarieswasreportedtoworkfullywithoutproblemswithQt4.Well,andwehadanedu-cationallicenseofQt3,so...but,asmentionedpreviously,thisisGPL’edsources-goaheadand...(pleaseseefollowingentryinthisFAQ)6.“Ihadsometimeduringtoday’slunchbreakandportedthesourcecodeofneur-ALCtoTrolltech’sQt4.DoyouliketoincludethechangestotheneurALCsourcetree?”-HOWGREAT!THANKS!!And:YES!Pleasecontactusbyemailorthroughhttp://neuralc.sourceforge.net7.“Howtoreference/quote/citeneurALC?”-IfyoupublishresultsinanarticleoruseneurALCasdataprocessorforanyotherkindofpublication,pleaseuseaphrasesimilarto:“Temporalresponseclasseswereobtainedfromtheanalyzedmulti-electroderecordingsusingtheneurALCprogram”inthemethodssection,andincludethefollowingtoyourlistofreferences:neurALC-across-platform,opensourceprogramfortheanalysisofextra-cellularneuronalmulti-electroderecordingsdistributedunderGPL(http://neuralc.sourceforge.net).

11Bibliography&References

M.Casdagli,Recurrenceplotsrevisited,PhysicaD108,206(1997)

G.CeleuxandG.Govaert,AClassificationEMalgorithmforclusteringandtwostochasticversions,ComputationalStatisticsandDataAnalysis,14(3):315-332.(1992)

J.P.Eckmann,S.OliffsonKamphorstandD.Ruelle,Recurrenceplotsofdynami-calsystems,Europhys.Lett.4,973(1987)

FFTW3-librarysystemforFourierandrelatedcalculations.→http://www.fftw.org

38

A.M.FraserandH.L.Swinney,Independentcoordinatesforstrangeattractorsfrommutualinformation,Phys.Rev.A33,1134(1986)

KlustaKwik-aprogramforunsupervisedclassificationofmulti-dimensionalcon-tinuousdata(seeKDHarrisetal,JournalofNeurophysiology84:401-414,2000)→http://klustakwik.sourceforge.net

MySQL-anopensourcedatabase.→http://www.mysql.com

NEV2lkit-apreprocessorforintra-andextra-cellularneuronalrecordingsdis-tributedunderGPL→http://nev2lkit.sourceforge.net

NEV-formatspecifications-asofthiswritingtheNeuralEventFormatspecifi-cations(version2.0)arepubliclyavailable.→http://cyberkineticsinc.com/NEVspc20.pdf

Qt-thecross-platformC++frameworkfromTrolltech.Pleasevisithttp://www.trolltech.com.

QWT-libraryfortheQtframeworkwhichprovidesQtwidgetsfortechnicalappli-cations.→http://qwt.sourceforge.net

QWT3D-libraryfortheQtframeworkwhichprovidesQtwidgetsforthree-dimensionalplotting.→qwt3dplot.sourceforge.net

TISEAN-asoftwareprojectfortheanalysisoftimeserieswithmethodsbasedonthetheoryofnonlineardeterministicdynamicalsystems.→http://www.mpiks-dresden.mpg.de/~tisean

12Licenses

Asmentionedthroughoutthismanual-neurALC,themanual,examplesanditssourcesarecoveredinfullorpartlybyoneofthefollowinglicenses.CopiesoftheGPLandLGPLcanbefoundinsidetheneurALCinstallationfolder.

GNUPublicLicense2.0(GPL)

ThefulltextoftheGPLisprovidedwithneurALCorcanbefoundathttp://opensource.org/licenses/gpl-license.php

GNULesserGeneralPublicLicenseVersion2.1(LGPL)

ThefulltextoftheGNULGPLisprovidedwithneurALCorcanbefoundathttp://opensource.org/licenses/lgpl-license.php

39

QWTLicense1.0

TheQwtlibraryandincludedprogramsareprovidedunderthetermsoftheGNULESSERGENERALPUBLICLICENSE(LGPL)withthefollowingexceptions:

1.WidgetsthataresubclassedfromQwtwidgetsdonotconstituteaderivativework.2.StaticlinkingofapplicationsandwidgetstotheQwtlibrarydoesnotconstituteaderivativeworkanddoesnotrequiretheauthortoprovidesourcecodefortheapplicationorwidget,usethesharedQwtlibraries,orlinktheirapplicationsorwidgetsagainstauser-suppliedversionofQwt.IfyoulinktheapplicationorwidgettoamodifiedversionofQwt,thenthechangestoQwtmustbeprovidedunderthetermsoftheLGPLinsections1,2,and4.3.YoudonothavetoprovideacopyoftheQwtlicensewithprogramsthatarelinkedtotheQwtlibrary,nordoyouhavetoidentifytheQwtlicenseinyourprogramordocumentationasrequiredbysection6oftheLGPL.However,programsmuststillidentifytheiruseofQwt.neurALCisbasedinpartontheworkoftheQwtproject(http://qwt.sf.net).

CreativeCommonsLicense(humanreadableform)

Thismanualisdistributedunderthefollowingacreativecommonslicense.ThefulltextisprovidedwithneurALCorcanbefoundat

http://creativecommons.org/licenses/by-nc-sa/2.0/

Attribution-NonCommercial-ShareAlike2.0Youarefree:

•tocopy,distribute,display,andperformthework•tomakederivativeworksUnderthefollowingconditions

AttributionYoumustgivetheoriginalauthorcredit.

NoncommercialYoumaynotusethisworkforcommercialpurposes.

Share-AlikeIfyoualter,transform,orbuilduponthiswork,youmaydistributethe

resultingworkonlyunderalicenseidenticaltothisone.

•Foranyreuseordistribution,youmustmakecleartoothersthelicensetermsofthiswork.•Anyoftheseconditionscanbewaivedifyougetpermissionfromthecopyrightholder.Yourfairuseandotherrightsareinnowayaffectedbytheabove.

40

ListofFigures

123456789101112131415161718192021222324252627282930313233343536373839

neurALCGUI:Population-tab.........................neurALCGUI:Electrodes-tab.........................neurALCGUI:3D-tab.............................neurALCGUI:Options-Tab..........................Options-Tab:Trigger..............................Options-Tab:Binning,ISI,Frequency.....................Options-Tab:Correlation............................Options-Tab:Delay...............................Options-Tab:Classification..........................Options-Tab:Selection.............................Options-Tab:“ElUltimo”............................Menubar:theneuralc-menu.........................Menubar:theFile-menu............................Menubar:theAnalysis-menuwiththeOverview-submenuselected...AnalysisMenu:theISI-submenu.......................AnalysisMenu:theCorrelation-submenu..................AnalysisMenu:theInstantfiring-menu....................Menubar:theContraptions-menu......................Contraptionsmenu:theConnecttodatabase-dialogue...........neurALC’ssimplifiedanalysisworkflow...................populationrasterplot..............................populationPSTHplot.............................exampleofaPSTH-basedISIdistributionofasingleelectrode......instantfiringratefunctionforthePSTHdataofasingleelectrode.....exampleofacrosscorrelationfunctionofasingleelectrode........delayrepresentationofapopulationrecording(delay=25ms,dimensions=2)defaultviewofabinnedcumulativePSTHvisualizedin3D.........defaultviewofaspectrogramcalculatedforthePSTHofaselectedelec-trode.......................................ContentofthefolderneuralcDB........................ExtendedEntityRelationDesignofthedatabaseusedbyneurALC...ThedbALCGUI................................RasterplotoffullexperimentaltimecourseandpopulationactivityoverlayZoomedpopulationactivity..........................Zoomedrecurrenceplotforthebinnedpopulationactivity..........Autocorrelationfunctionofthebinnedpopulationactivity...........Post-stimulustimehistogramforElectrode1withRate-overlay......ISIdistributionfortheselectedneuronal“unit”................CrosscorrelationfunctionforElectrode1,Unit3withpopulationactivity.ExampleASCII-exportforaselectedgroupofneuronal“units”......

567899101011111213131415151617171820202121222324242728293131323233343437

41

Contents

1SomeSmallPrint2Preface3Installation

3.1Linux....3.2MacOSX.3.3Windows..........................................................................................................2333443.4Compilation

...................................4Usage&Functionality

4.1TheneurALCGUI.......

........................4.1.1ThePopulation-Tab..........................4.1.2TheElectrodes-Tab..........................4.1.3The3D-Tab...............................4.1.4TheOptions-Tab............................4.2TheneurALCMenuBar....

........................4.2.1Theneuralc-menu...........................4.2.2TheFile-menu.............................4.2.3TheAnalysis-menu...........................4.2.4TheContraptions-menu........................4.2.5TheHelp-menu.....

........................5Analysis&VisualizationBackground

5.1neurALC’sAnalysisWorkflow.........................5.2SpikeTrainRepresentation..........................5.3MultivariateTimeSeriesAnalysisMethods.........

........5.3.1Correlation...............................5.3.2Recurrence...............................5.3.3Delay..................................5.3.4MutualInformation...........................5.43-DVisualization................................5.5Classificationofelectrode&“unit”responses.........

........5.5.1ConceptofTemporalResponseTypeClasses(TRC)........6MySQLDatabaseIntegration

6.1TechnicalPrerequisites............................6.2DatabaseDesign................................6.3dbALC-adatabaseGUI...........................6.4SQLcommandlogforcreatingtheneurALCdatabase...

........

7ARealWorldExample

42

45556781213131416171818192222222323242525262627283031

8Appendix

358.1KeyboardShortCuts..............................358.2MouseFunctionality..............................368.3SupportedFileFormats

............................368.3.1Input...................................368.3.2Output..................................

369Bugs

3710FrequentlyaskedQuestions3711Bibliography&References3812Licenses39ListofFigures

41

Colophon

ThismanualwaswrittenunderMacOSXandlayoutedfortwo-sidedprintingonA4(portrait)usingTeXShop1.40andteTEX.TheneurALCanddbALCandsomeofthemarginaliconswerecreatedbyM.BongardusingAdobeIllustratorandLemkesoft’sGraphicConverter.ScreenshotsweretakenunderMacOSX10.3.9and10.4.2andprocessedusingApple’sMacOSXpreviewapplication.

43