Gene2Net User Manual October 23, 2015 Gene2Net (www

Gene2NetUserManual
October23,2015
Gene2Net(www.gene2net.org)isaweb-basedapplication,whichallowsbiologists
toexpandoneormultiplegenesintoasmallnetworkbasedonthethreehighly
functionallyrelevantproteinco-expressionnetworksgeneratedfrombreastcancer,
colorectalcancerandovariancancerproteomicsdatasets,performs“guilt-byassociation”functionalanalysisbasedontheGeneOntologydatabase
(www.geneontology.org)andcomparesdifferentsub-networksandcorresponding
functions.Theusercanfindthenodeandedgeinformationaboutthreeproteincoexpressionnetworksontheintroductionpage(Figure1).Clicking“START”button
ontheintroductionpage,theusercanstarttoexecuteananalysisontheGene2Net.
Figure1.IntroductionpageofGene2Net.
1.Desiredoptionselectionandseedinput
a.Cancertypeoption
Gene2Netgivestheoptionofchoosingoneormultiplecancertypesforanalysis.The
usermaychoosethecancertype(s)byclickingthedropdownmenuunder“Selected
cancertype”(Figure2a).Andthentheco-expressionnetwork(s)correspondingto
thecancertype(s)willbeusedforthefollowinganalysis.
b.Networkconstructionoption
Oncethecancertypesareselected,theusercanchoosethenetworkconstruction
methodforconstructingthesub-networksrelatedtotheinputseeds(Figure2b).
ThetwooptionsforthenetworkconstructionmethodareNetworkExpansionand
NetworkRetrieval.NetworkExpansionmethodusesrandomwalkanalysistorank
allthegenesintheco-expressionnetworkbasedontheirnetworkproximitytothe
inputseeds.Itthenreturnsasub-networkinwhichnodesaretheinputseedsand
theirtoprankingneighborsandedgesrepresenttheirco-expressionrelationships.
NetworkRetrievalmethodidentifiesco-expressionrelationshipsbetweentheinput
seedsinthenetworkandreturnsasub-networkinwhichnodesaretheinputseeds
andedgesrepresenttheirco-expressionrelationships.IfchoosingNetwork
Expansion,theuserneedstospecifythenumberoftoprankingneighborstobe
displayedontheoutput.Thenumberondefaultissetto20.(Figure2b).
c.Significantlevelmethodoptionforenrichmentanalysis
TwomethodsareavailabletotheuserforidentifyingsignificantGeneOntology(GO)
terms:FDRandTop(Figure2c).Gene2Netfirstcalculatesthepvaluesof
enrichmentforallGObiologicalprocesstermsforgenesintheidentifiedsubnetwork.AfteradjustingthepvaluesbasedontheBenjamini-Hochbergmultiple
hypothesiscorrection,FDRmethodwillidentifythesignificantGOtermsbasedon
theinputFDRthresholdwhileTopmethodwillranktheGOtermsfromlowest
adjustedpvaluetohighestadjustedpvalueandselectthetopnumberofGOterms
assignificantterms.TheFDRthresholdisondefaultsetto0.05andtheTopnumber
ofGOtermsisondefaultsetto10.
d.Highlightoption
Becausetheusermaywanttohighlighteithertheseedsortheneighborsofthe
outputgraphgeneratedbytheNetworkExpansionmethod,theusercanchoose
whichonetohighlightbyclickingontheappropriateradiobuttoninFigure2d.
a
c
b
d
e
Figure2.Desiredoptionselectionandseedinput.a.Desirednetworkselection.b.Network
constructionmethodselection.c.Significantlevelmethodselectionforenrichmentanalysis.d.
Highlightselection.e.Seedinput.
e.Seedinput
Afterselectingthedesiredsettings,theusercaninputtheseedsdirectlyintothe
textboxunderneaththesettings.Iftheuserselectsonecancertype,suggestions
appearforgenesinthecorrespondingco-expressionnetworkoncetheusertypesat
leastoneletterintothetextbox.Ifselectingmultiplecancertypes,suggestionswill
includethecommongenesamongallco-expressionnetworksofselectedcancer
types.Theusermayinputasmanyseedsasdesiredandcanremovetheseedsfrom
theinputarraybyselectingthe“X”bytheleftofthegene(Figure2e).Theusermay
alsocopy-pastealistofseedsintothetestbox.Thelistofseedsmaybespace
separated,tabseparated,orlineseparated.Sincetheinputseedsmaynotbe
includedintheco-expressionnetwork(s)ofselectedcancertype(s),thistypeof
geneswillbeshowninthe“Invalidinputs”textbox(Figure2e).
Afterselectingoptionsandinputtingseeds,theusermaypressthesubmitbutton
underneaththegenesymboltextboxtostarttheanalysis.Theanalysiswilltakea
fewsecondsuptoaminute.
2.Resultvisualization
Theanalysisresultsofeachcancertypeincludetheidentifiedsub-networkandthe
correspondingenrichedGOterms.AsshowninFigure3,thesub-networkwillbe
visualizedbythecytoscapewebtool(http://cytoscapeweb.cytoscape.org/)and
enrichedGOtermswillbeorganizedbasedonGODAG(directedacyclicgraph)
structureandvisualizedbythetooldevelopedbasedonthextracepackage
(https://github.com/brownsys/xtrace).Gene2Netalsoprovidesthedetail
informationfornodesinthesub-networkandstatisticscoresofenrichedGOterms.
Allresultsofeachcancerwillbeorganizedinatabpage.Iftheuserselectsmultiple
cancertypes,Gene2Netwillcomparetheresultsfromeachcancertypeandgenerate
twomoretabpages,“UnionGraph”and“IntersectGraph”.Detailedinformation
aboutthesetabscanbefoundbelow.Theusercanclickthe“DownloadData”button
or“NewAnalysis”buttonatthetopofthepagetodownloadallanalysisresultsor
startanewanalysis.
Figure3.Resultvisualization.
a.Sub-networkvisualization
Thesub-networkgraphisvisualizedontheupperleftofeachtabpage.Iftheuser
selectsNetworkExpansionmethod,thegraphwilldisplaytheinputseedsaswellas
thetoprankingneighbors.Thesizeofthenodeandnodelabelwillbedetermined
bytheselectionofthe“Highlight”optionintheoptionspage.Iftheuserselectsto
highlighttheseeds,theseedswillhavealargernodeandnodelabelinthegraph
comparedtotheneighbornodes(seeFigure3).FortheNetworkRetrievalmethod,
thegraphwillonlydisplaytheinputseedsandthusnodeandlabelsizearethesame
betweennodes.Theusermaydownloadthegraphunderavarietyofoptionsby
choosingthedesiredextensionfordownloadunderneaththe“Exportas:”inthe
dropdownmenu,thenclickingonthe“Export”button.
b.EnrichedGOtermvisualization
TheDAGstructureofenrichedGOtermsisontheupperrightofeachtabpage
(Figure3).Redboxesrepresenttheenrichedtermswhileyellowboxesrepresent
theancestorsoftheseterms.Theusermayzoominandoutthroughusingthe
mousewheel.ClickingasignificanttermwillzoomtheDAGintotheareaaroundthe
termandalterthebordercoloroftheselectedtermtoblue.Meanwhile,thegenes
annotatedtotheselectedtermwillalsobehighlightedinthesub-networkgraphby
changingtheirnodecolortobered(Figure3).The“EnrichedGOTerms”tableinthe
bottomrightofthepagewillalsobeautomaticallyscrolledtotherowthatcontain
thedetailedinformationoftheselectedterm(Figure3).Theusermaymovethe
viewportbyholdingtheleftmousebuttononthegraphanddraggingtothedesired
location.Thecurrentlocationoftheviewportofthegraphcanbeseeninthe
minimaponthebottomrightsideofthegraph.Similartothemaingraph,the
minimapcanbezoomedanddraggedtochangetheviewportofthegraph.
Onhoveringoveraterm,theterminformationwillbeshowninapop-updialogand
itsancestorsanddescendantswillbehighlighted(Figure4).Theusermayselect
multipletermthroughfirstclickingonaterm,thenholdingtheshiftbuttonand
clickinganotherterm,whichselectsallofthetermsinthepathbetweenthetwo
terms,ifthereisapath,orselectbothnodesifthereisnotone.Similarly,the
controlkeycanbeusedinWindowscomputerstoselectmultiplenodes.Theuser
mayalsorightclickonanodeandgetmultipleoptions,suchaslookingupthe
terms’detailedinformationontheGeneOntologywebsite,hidingthecurrently
selectednodes,non-selectednodes,invertingtheselection,selectingall,selecting
theadjacentneighborsandthepath(Figure5).Uponclickinghidenodes,thegraph
redrawsitselfwithouttheselectednodes.Toreturnthetermsbackandtounhide
theterms,clickonthe“UserSelection”onthebartotheleftofthegraphandthe
termswillbeunhidden(Figure6).
Underneaththegraph,therearethreebuttonstoimprovetheusabilityofthegraph.
ThefirstbuttonallowstheusertodownloadthecurrentonscreenGOgraph–todo
soclickthedownloadbuttonandthenclickthedesiredextensionfromthe
dropdownmenu.ThesecondbuttonallowstheusertosearchforaGOterminthe
graph.Onclick,thedropdownmenuopensandallowstheusertotypeterms.There
isanauto-fillfunctionattachedwiththesearchbox:typingpartofeithertheGO
termIDortheGOfunctionnamewillallowtheusertoseealistofsuggestionson
whattosearch.Oncethetermisselected,theusercanpressthesearchbuttontothe
rightofthetextboxinordertozoomintothetermintheGOgraph(Figure7).
Finally,onclickofthefinalrefitbuttontotheright,theGOgraphrefitstothe
maximumallottedspacegiventothegraph.
Figure4.Highlightfunctionwhenhoveringonaterm.
Figure5.Moreoptionswhenright-clickingthegraph
Figure6.Thefunctionforunhidingtheterms.
Figure7.SearchfunctionforGOtermsintheEnrichedGOtermgraph.
c.Tablesfordetailedinformation
Thebottomofthespecificcancerpagecontainsdetailedinformationaboutthe
analysis,asseeninFigure3.Thefirsttableontheleftcontainstheinputseeds.The
seedsthatareinthenetworkareshowninablackfontwhereastheseedsthatare
notinthenetworkareshowninabluefont.Theusermayclickonthegeneto
redirecttheusertotheNCBIpageonthegene.Thesecondtablecontainsthetop
rankingneighbors(notavailableforNetworkRetrievalmethod).Thegenesymbols
andrandomwalkprobabilityareshown.Theusermay,again,clickonthegene
symboltoseethedetailedinformationontheNCBIpage.Finally,thelasttable
containstheenrichedGOTermsoftheidentifiedsub-network.Thetablecontainsa
checkbox,whichcanbemanuallycheckedtochangethecolorofthegenes
annotatedtotheselectedtermasredinthesub-networkandalsoscrolltothenode
ontheGOGraph.TheusermayclickontheGOIDtosearchtheGeneOntologyterm
intheGeneOntologywebsite.Therelatedseedsandneighborsarealsoshowninthe
table.ForNetworkRetrievalmethod,iftheuserselectstohighlighttheseeds,the
seedsinthetablewillhaveabiggerredfontwhileneighborswillhaveabiggerred
fontiftheuserselectstohighlighttheneighbors.
d.Multiplecancertypecomparison
Iftheuserselectsmultiplecancertypesinthecancertypedropdownmenu,
Gene2Netnotonlyvisualizestheanalysisresultsforeachcancertypebutalso
comparetheresultsamongcancertypes.AsshowninFigure3,usermayclickonthe
UnionGraphtabtogetacomparisonresultsforthecombinationofallsub-networks
fromdifferentcancertypesorclickontheIntersectGraphtabtogetanalysisresults
fortheintersectgenesamongallsub-networks.
AsshowninFigure8,thetopsectionoftheUnionGraphisanetworkcomparison.
Totheleftisacombinedgraphcontainingthesub-networksfrommultiplecancer
types.Theedgesofdifferentsub-networksislabeledbydifferentcolors.Ifanedge
hasappearedinmultiplesub-networks,theedgewillbeplottedmultipletimeswith
differentcolorsinthecombinedgraph.TotherightofthecombinedgraphisaVenn
diagramforthecomparisonamongsub-networksofdifferentcancertypes.
NumbersarelabeledeachVenndiagramportion,depictingthenumberofgenesthat
areuniqueorsharedbetweenthesub-networks.Theusermayclickonthe
intersectionordifferencebetweenthesetsandgenesinthissectionwillbe
highlightedintheleftcombinedgraphbychangingthebordercolorofthesegenes
toblue.Theusermaychoosetoresetthebordercolorbyclickingthebeigearea
outsidetheVenndiagram.
ThebottomsectionoftheUnionGraphisthecomparisonbetweentheenrichedGO
terms.AsshowninFigure8,theleftoftheUnionGraphisaGODAGstructure
containingallsignificantGOtermsfromthedifferentsub-networks(redcolor)and
theirancestorterms(yellowcolor).Similartothenetworkcomparison,function
comparisonalsousesaVenndiagramtoillustratethesimilaritiesanddifferences
betweentheenrichedfunctions.ClickingasectionontheVenndiagramwillchange
theGOtermsinthissectionasadarkbluecolorintheleftGODAGgraphwhilethe
redcolorofallothersignificanttermswillbedesaturated.Clickingonthesection
willalsogivetheuseralistoftheGOIDsandGOnamesofthetermsinatable.
ClickingonanyredorblueGOtermsintheleftGOgraphcanchangethecolorof
genesannotatedtotheselectedtermasredinthecombinedgraphofNetwork
Comparisonsection.
ThefinaltabinFigure9depictstheIntersectgraph,whichextractsthecommon
genesamongthesub-networksfromtheselectedcancertypesandperformsthe
enrichmentanalysisbasedonthecommongenes.Thecolorsofedgesbetween
commongenesarethesameasthoseintheUniongraph.Otherfunctionsinthe
Intersectgrapharethesameasthoseinthetabpageofindividualcancertype.
Figure8.ComparisonresultsinUniongraph
Figure9.AnalysisresultsinIntersectgraph