Original Article · 2016. 1. 31. · Chemoarchitecture Chemoarchitecture reports as collated from...

Neuroinformaticscopy Copyright 2006 by Humana Press IncAll rights of any nature whatsoever are reserved1539-279106275ndash298$3000 (Online) 1559-0089DOI 101385NI44275

Original Article

A New Module for On-Line Manipulation and Displayof Molecular Information in the Brain ArchitectureManagement SystemMihail Bota and Larry W Swanson

The Neuroscience Research Institute University of Southern California 3641 Watt Way Los AngelesCalifornia 90089-2520 USA

275

Author to whom all correspondence and reprint requests should be addressed E-mail mbotaalmaakuscedu

AbstractA new ldquoMoleculesrdquo module of the Brain

Architecture Management System (BAMShttpbrancusiuscedubkms) is describedWith this module BAMS becomes the first on-line knowledge management system to handlecentral nervous system (CNS) region and cell-type chemoarchitectonic data in the context ofaxonal connections between regions and celltypes in multiple species The ldquoMoleculesrdquomodule implements a general knowledge rep-resentation schema for data and metadatacollated from published and unpublishedmaterial and allows insertion of complexreports about the presence of moleculescollated from the literature For different CNSneural regions and cell types the modulersquosdatabase structure includes representation ofmolecule expression revealed by various tech-niques including in situ hybridization andimmunohistochemistry molecule coexpres-sion and time-dependent level changes and

physiological state of subjects The metadatarepresentation allows online comparison andevaluation of inserted experiments andldquoMoleculesrdquostructure allows rapid develop-ment of data transfer protocols enabling neu-roinformatics visualization tools to displaygene expression patterns residing in BAMS interms of levels of expressed molecules andin situ hybridization data The modulersquos webinterface allows users to construct lists of CNSregions containing a molecule (depending onphysiological state) retrieve further details aboutinserted records compare time-dependent datawithin and across experiments reconstructgene expression patterns and construct com-plex reports from individual experiments

Index Entries Chemoarchitecture coexpres-sion database immunohistochemistry infer-ence engine in situ hybridization moleculesmolecular neuroanatomy

(Neuroinformatics DOI 101385NI44275)

276 __________________________________________________________________________Bota and Swanson

Neuroinformatics_________________________________________________________________ Volume 4 2006

Introduction

The vast amount of neuroscience-related tex-tual and graphical informationmdashwhich actu-ally extends back centuriesmdashis one importantfactor that makes searching and interpretationa difficult task In addition the interpretationand integration of data are derived from dif-ferent levels of central nervous system (CNS)organization Any CNS region can be describedwith respect to multiple levels of organizationfrom patterns of gene expression under spe-cific experimental conditions to structuralchemical and physiological characteristics ofits constituent neuronal and glial cell types tospecific roles in functional networks of regionsin a particular species (Swanson 2000ab Botaet al 2003)

Database systems are complementary to theneuroinformatics visualization tools and appli-cations developed by individual researchgroups and are necessary for data storage andhandling The display of multiple modalitiesmdashincluding structural and functional data andgene expression patterns (levels of expressedmolecules and in situ hybridization data)mdashonthree-dimensional brain reconstructions ortwo-dimensional atlas reconstructions ofbrains from various species requires the designand development of complex backend data-bases as information repositories

To address these problems we are develop-ing the on-line knowledge management systemBrain Architecture Management System(BAMS) for systematizing organizing and pro-cessing neuroscience information relevant todifferent levels of CNS organization The struc-ture of BAMS allows insertion of data about CNSregions cell types neural (axonal) pathwaysand molecules collated from the literature orrecorded by users BAMS includes inferenceengines that establish qualitative topologicalrelations between CNS regions reconstruct pro-jection patterns from CNS regions of interestand construct possible networks of CNS regions

from connectivity data resident in the system(Bota et al 2005)

BAMS is also a data provider for neuroinfor-matics systems developed by other groups Forexample simple data transfer protocols havebeen implemented with several web-based andstand-alone applications created by theLaboratory of NeuroImaging (LONIhttpwwwloniuclaedu) at UCLA that provide hierarchically organized neuroanatomicalnomenclatures and definitions of CNS parts(MacKenzie-Graham et al 2003 2004)

Here we describe the structure of a newBAMS module ldquoMoleculesrdquo that allows inser-tion of molecular (chemical) data pertaining tovarious CNS regions and cell types We alsodescribe here the user interfaces designed toallow searching for molecules in BAMS whosedatabase structure and inference engines aredescribed elsewhere (Bota et al 2005)

Methods

BAMS is hosted on a Dell 8200 desktop com-puter with a Pentium III processor runningunder the Windows XP operating system withIIS 51 The backend relational database of BAMSwas created in MySQL and PHP was used toconstruct web interfaces and inference enginesThe architecture of BAMS is constructed on threelevels a set of tables constructed in MySQL thatstore data collated from the literature (or insertedby neuroanatomists) an intermediate levelencoded in PHP that includes queries and algo-rithms for processing data populating BAMSand an output level that is mainly in an HTMLtabular format and graphics

ldquoMoleculesrdquo Module Database Structure

The BAMSrsquos ldquoMoleculesrdquo module consistsof two parts a public part and a ldquoPersonalaccountrdquo part that was designed to store dataentered by collators and restricted to their useThe knowledge representation schema ofBAMSrsquos ldquoMoleculesrdquo module (Fig 1) follows

Volume 4 2006_________________________________________________________________ Neuroinformatics

A New Module for On-Line Manipulation and Display of Molecular Information in the BAMS ___________277

the general organization and presentationof experimental data in published neuroscienceresearch articles

Any reference inserted in BAMSrsquosldquoMoleculesrdquo module can be associated withmultiple experiments An experiment is definedas an experimental paradigm applied to a groupof animals or human subjects Each experimentmay consist of several experimental serieswhich are defined by the specific procedures(eg different antibodies or nucleic acid probes)that were applied Each experimental series isassociated with a set of experimental data thatinclude mapped brain regions and measuredvariables The conceptual design of BAMSrsquosldquoMoleculesrdquo module completely separatesexperimental data from metadata collated fromthe reference or inserted by collators Metadataclasses associated with data and experimentsinserted in BAMSrsquos ldquoMoleculesrdquo moduleinclude mapping experimental series experi-ment data presentation and physiological stateEach of these metadata classes is described later

The general object-relationship schema thatstores data and physiological state metadata inBAMSrsquos ldquoMoleculesrdquo module is presented inFig 2 Each object and relation shown in thefigure may be captured in more than one tableThe backend MySQL relational database of theldquoMoleculesrdquo module public part now has 17tables (listed in Table 1)

The table Brain part contains the unique iden-tification of CNS parts (gray matter fiber tractsor ventricles) as collated from neuroanatomical

nomenclatures and it was described in Bota et al(2005) The table Chemoarchitecture allows inser-tion of a large range of experimental data fromqualitative assessments of the presence orabsence of molecules in different CNS neuralregions to quantitative measurements of suchdata Qualitative attributes that describe the dis-tribution of a molecule in a CNS region includestaining strength staining pattern and topo-graphical position of stained cells in the regionThe assessment of qualitative strengths is simi-lar to that for neuroanatomical projectionsbetween regions as described in detail elsewhere(Bota et al 2005) Chemoarchitecture also allowsinsertion of data pertaining to cell counts per-cent of cells in terms of total area or per area andaverage number of labeled cells (plusmnstandard devi-ation or standard error of the mean) Quantitativeand semiquantitative data that can be insertedin this table include optical absorption absorp-tion level relative to background standard devi-ation and standard error of the mean

The ldquoMoleculesrdquo module also allows inser-tion of qualitative and quantitative reports ofmolecule expression in different cell types asrevealed by ligand autoradiography immuno-cytochemistry or in situ hybridization BAMSrsquosldquoMoleculesrdquo module includes the tableCoexpression that contains reports of moleculecoexpression in CNS regions or cell typesunder various physiological conditions as col-lated from the literature The presence orabsence of a molecule in a CNS region or celltype often depends on the age sex andor

Fig 1The knowledge representation schema of ldquoMoleculesrdquo module in BAMS

278 __________________________________________________________________________Bota and Swanson


physiological state of the subjects Thereforeany database that stores data related to thepresence of molecules in different CNS regionsshould include physiological state metadata(see Fig 2) The ldquoMoleculesrdquo module includesa simple representation of the subjectrsquos phys-iological state encoded in the following tablesPhysiological_state Manipulation_type andManipulation_data The table Physiological_stateencodes for the values allowed in BAMS whichare ldquonormalrdquo and ldquomanipulatedrdquo An experi-mental animal is considered to be in a ldquonor-malrdquo physiological state when the authors ofthe collated reference state that no special

procedure was applied The physiological stateldquomanipulatedrdquo currently refers either to injec-tions of chemicals or to structural lesions indifferent CNS regions or parts of the experi-mental animal The structural lesions pro-duced by different chemicals injected in theCNS are classified as chemical manipulationsTherefore the physiological state ldquomanipu-latedrdquo is represented in BAMS by two possi-ble values ldquochemical manipulationrdquo andldquostructural manipulationrdquo which are encodedin the table Manipulation_type The tableManipulation_data stores details about thelesioned CNS parts the side of the brain where

Fig 2 The object-relationship schema of Moleculesrsquo public part for experimental data and physiological state metadata



the procedure was performed the injectedchemicals or lesioned CNS regions and anno-tations from the collated references

The BAMS ldquoMoleculesrdquo module also includesseveral tables that allow insertion of time-dependent molecular data The range of experi-mental data and variables that can be insertedin the tables time_dependence-chemoarchitectureand time_dependence-coexpression are identicalwith those in chemoarchitecture and coexpressionrespectively Records inserted in either of thesetables are grouped in experiments as reportedin collated references This allows recordingthe statistical significance of experimental datacalculated within or across experiments asreported in collated references

ldquoMoleculesrdquo Module Classification Schema

In this initial implementation ldquoMoleculesrdquoincludes a simple classification schema for itsnamesake with just two general classes of mol-ecules ldquoreleasable by neuronsrdquo and ldquocell asso-ciatedrdquo which are encoded in the tableMolecule_type These two classes partition com-pletely the set of molecules that can be insertedin BAMS because any chemical that can beidentified in a CNS region is considered herefor simplicity to be produced by neurons (glialcells will be incorporated in the future) Againfor simplicity the current partition of mole-cules in BAMS is of the disjoint type That isany molecule can be classified in either of the

Table 1The Tables Included in the Public Part of BAMSrsquos ldquoMoleculesrdquo Module That Represent Experimental

Data and Physiological State Metadata

Table in BAMS Encodes for

Brain part Unique identification and basic information about brain parts collated from different neuroanatomical atlases or nomenclatures

Molecule Name and abbreviation of those chemicals inserted in BAMSMolecule_type The major classes of molecules allowed in BAMS cell associated or

releasable by neuronsMolecule_releasable Subclasses of molecules considered in BAMS to be releasable by neuronsCell_associated Subclasses of molecules considered in BAMS to be found in cells and not

released by neuronsEnzymes Basic information about enzymes inserted in BAMSReceptors Basic information about receptors inserted in BAMSReceptors subunits Basic information about subunits of receptors inserted in BAMSChemoarchitecture Chemoarchitecture reports as collated from the literatureCoexpression Reports of molecule coexpression identified in different brain regions or cell

types and collated from the literatureTime dependence- Reports of time-dependent chemical data

chemoarchitectureTime dependence- Reports of time-dependent coexpression data

coexpressionStatistical significance Information about statistically relevant data measured across or within

experimentsPhysiological state Allowed values of the subjectrsquos physiological stateManipulation type Information about the chemical or structural manipulations allowed

in BAMSManipulation data Information such as location hemisphere and associated annotation about

the physical or chemical manipulation performed on subjects

280 __________________________________________________________________________Bota and Swanson


general classes but cannot be an instance ofboth The design of the ldquoMoleculesrdquo moduleallows each of these two classes to be furtherdivided into subclasses The class ldquocell associ-atedrdquo is subdivided at present into two sub-classes ldquoenzymesrdquo and ldquoreceptorsrdquo The classldquoreceptorsrdquo is further subdivided into ldquorecep-tor subunitsrdquo Each of these classes and sub-classes is assigned to a table that encodes forrelationships between parents and childrenand names and abbreviations of the classifiedmolecules The classification schema includedin ldquoMoleculesrdquo allows new children to beinserted into either of the most general classes

Metadata

The class ldquoExperimental series metadatardquoshown in Fig 1 allows description of proce-dures applied to series included in an experi-ment and is divided in three subclasses thatcorrespond to the techniques encoded inBAMSrsquos ldquoMoleculesrdquo module ligand autora-diography immunohistochemistry and in situhybridization The fields that make each ofthese subclasses are listed in Table 2

Ligand autoradiography metadata describereceptor mapping experiments using radioac-tively labeled ligands In constructing this meta-data subclass we followed Burt (1985) Kuhar(1985) and Kuhar et al (1986) The ldquoequilibriumdissociation constantrdquo and ldquomaximum specificbindingrdquo fields are used to describe ligand-receptor interactions and measure the ligandaffinity and the maximal number of boundreceptors respectively The ldquoquantitation stan-dardrdquo and ldquoquantitation methodrdquo fields describethe procedures used to quantify receptor autora-diograms (Burt 1985 Kuhar et al 1986)

Immunohistochemistry metadata describeexperiments using antibodies to reveal themolecular phenotypes of neurons Fieldsincluded in this metadata class subtype havebeen discussed in Watts and Swanson (1989)Burry (2000) Saper and Sawchenko (2003) andSaper (2005) The field ldquoperformed controlrdquo

specifies what type of control was performedto ensure antibody specificity and allowed val-ues are ldquoknockoutrdquo ldquoin situ hybridizationrdquoldquoimmunoblotrdquo ldquoimmunoprecipitationrdquoldquoadsorptionrdquo ldquostaining patternrdquo and ldquonotspecifiedrdquo (Saper 2005)

In situ hybridization metadata describe neu-ronal molecular phenotype mapping withnucleic acid probes The fields included in thismetadata class subtype are used extensively inthe literature (see for eg Swanson andSimmons 1989 Wada et al 1989 Watts andSwanson 1989 Watts and Sanchez-Watts2002) The ldquolabeling methodrdquo field encodes forthe technique used to label the probe sequenceand the allowed values are ldquoradiolabeledrdquoldquoantigen labelingrdquo ldquofluorescencerdquo and ldquonotspecifiedrdquo

All three subclasses of ldquoExperimental seriesmetadatardquo have several common fields ldquosource(provider)rdquo ldquovisualization methodrdquo ldquovisual-ization supportrdquo and ldquoannotationrdquo Theldquosource (provider)rdquo field can be used to regis-ter the company or laboratory that providedthe ligand antiserum or nucleic acid sequenceused This information can be very importantin determining the specificity and affinity of aspecific ligand or antiserum (Kuhar et al 1986Sawchenko and Saper 2003 Saper 2005) aswell as for other specific information Theldquovisualization methodrdquo field specifies proce-dures used to visualize target molecules andallowed values are ldquoautoradiogramrdquo ldquofluo-rescencerdquo ldquoPETrdquo ldquocolorimetryrdquo and ldquonot spec-ifiedrdquo Allowed values for the ldquovisualizationsupportrdquo field are ldquofilmrdquo ldquosliderdquo ldquononerdquo andldquonot specifiedrdquo Finally the ldquoannotationrdquo fieldcan be used to insert more details about eachof the techniques

The metadata associated with an experiment(see Fig 1) are listed in Table 3 This metadataclass is made up of three subclasses andincludes two fields that specify the method ofneuron identification and the staining fre-quency respectively The ldquoAnimals (subjects)rdquo



subclass allows insertion of basic informationabout the group of animals or subjects includ-ing number sex weight age and living con-ditions The ldquoEmployed techniquerdquo subclasslists the experimental techniques used in theassociated experiment The ldquoAnatomical meta-datardquo subclass allows insertion of details asso-ciated with tissue preparation

Finally metadata classes associated with a ref-erence in BAMSrsquos ldquoMoleculesrdquo module ldquoDatapresentationrdquo and ldquoMappingnomenclaturemetadatardquo are listed in Table 4 The class ldquoDatapresentation metadatardquo includes three sub-classes ldquoCoordinatesrdquo ldquoDescription typesrdquo andldquoData typesrdquo The subclass ldquoCoordinatesrdquo allowscollators to insert metadata related to sterwotaxic

Table 2 Metadata Subclasses and Fields Included in the ldquoExperimental Series Metadatardquo Class

Experimental series metadata Ligand autoradiography LigandLigand source (provider)Equilibrum dissociation

constant (KD)Maximum specific binding

(Bmax)Quantitation standardQuantitation methodVisualization methodVisualization supportAnnotation

Immunohistochemistry Antibody source (provider)AntigenAntigen speciesPrimary antibodyPrimary antibody speciesPrimary antibody typeSecondary antibodySecondary antibody

speciesImmunoglobulin classPerformed controlVisualization methodVisualization supportAnnotation

In situ hybridization Sequence type mRNAheterogeneous nuclear (hn)RNA

Sequence source (provider)SequenceSequence directionSpecies sequenceControl directionLabeling methodVisualization methodVisualization supportAnnotation

282 __________________________________________________________________________Bota and Swanson


or Talairach coordinates associated with dataThe ldquoDescription typesrdquo subclass is similar tothe precision description codes implemented inCoCoMac (Stephan et al 2001) and its fieldsinclude possible ways of presenting data in a reference from textual descriptions to complete sets of maps representing labeled CNS regionson templates of reference or standard neu-roanatomical atlases The ldquoMappingrdquo metadataclass allows collators to specify how a recordabstracted from a reference was associated withCNS part names from nomenclatures insertedin BAMS This metadata class is important formaintaining the accuracy and consistency of datain BAMS (see ldquoCuration and Insertion of Datardquobelow) The ldquoData typesrdquo metadata subclass con-tains types of data associated with the referencequantitative semiquantitative or qualitativeThis metadata subclass is not inserted by colla-tors but is inferred by the system from insertedinformation

Curation and Insertion of Data

One of the most important problems in pop-ulating and maintaining a database is the

quality of inserted data To help ensure accu-racy of data entry in BAMS we designed amultistep comprehensive curation policythat is applied to all information inserted inpublic modules This policy involves differ-ent actions by BAMSrsquos collators and curatorsas well as by authors of inserted referencesThe curation policy schema implemented inBAMS is shown in Fig 3

The first step of BAMSrsquos curation method-ology is related to data sources All informa-tion inserted in BAMSrsquos public modules iscollated from published literature and is doc-umented by referencesmdashdata inserted inBAMSrsquos public modules is collated from orig-inal research articles Methods for handlingthe literature and BAMS database tables thathold information about inserted references aredescribed in Bota et al (2005) and Bota andArbib (2004) The second step in BAMSrsquos cura-tion methodology is data mapping onto one ofthe inserted neuroanatomical nomenclaturesAny record inserted in BAMS must be associ-ated with one of its neuroanatomical nomen-clatures Allowed mappings of a statement in

Table 3 Subclasses and Fields Included in the ldquoExperiment Metadatardquo Class

Experiment metadata Animals (subjects) Number SexWeightAgeLiving conditionsAnnotation

Neuronglia identification methodStaining frequency (sampling)Used technique Ligand autoradiography

ImmunohistochemistryIn situ hybridization

Anatomical metadata Sectioning methodSection orientationAngle of cuttingSection thicknessPreservation methodStaining methodStaining frequencyAnnotation



the general form ldquomolecule X is present in CNSregion Yrdquomdashin one of BAMSrsquos nomenclaturesmdashare listed in the ldquoMappingnomenclaturerdquometadata class The first of two currentlyallowed mappings is based on a statement inthe collated reference that refers explicitly to the neuroanatomical nomenclature usedHere the collator simply checks whether thenomenclature is inserted in BAMS If it isinserted and the collator wishes to use thisnomenclature no further action is necessaryThe second allowed mapping occurs when thecollator wishes to map all or part of the dataonto a different nomenclature resident inBAMS This remapping can be performed by

the collator andor original author with theprocedure stated clearly in the annotation asso-ciated with all records in BAMS If neither ofthese mapping possibilities is met the recordcannot be inserted in BAMS To circumvent thiscondition the collator (or original author) maywish formally to enter a new satisfactorynomenclature into BAMS so that the secondallowed mapping procedure can be followed

In short BAMSrsquos curation policy has two rulesfor data entry in its public modules data is col-lated from published original research articlesand each record has to be mapped onto one ofBAMSrsquos nomenclatures Hence association withthe ldquoMapping metadatardquo class is mandatory for

Table 4Subclasses and Fields Included in ldquoData Presentationrdquo and ldquoMappingNomenclaturerdquo Metadata Classes

Data presentation metadata Coordinates LambdaBregmaInteraural distanceX TalairachY TalairachZ Talairach

Description types TextRepresentative labeled imagesIntegral set of labeled imagesRepresentative labeled images

with drawn boundariesIntegral set of labeled images

with drawn boundariesRepresentative drawings or

mapping on atlas templatesIntegral set of drawings or

mappings on atlas templatesData types Quantitative data

Semiquantitative data (ratios)Qualitative data

Mapping metadata Brain region is captured in original published nomenclature in BAMS

Brain region is captured in a BAMS nomenclature other thanthat used in the original publication

Brain region is not in a BAMS nomenclature and the mappingto a BAMS nomenclature was performed by the author

Brain region is not in a BAMS nomenclature and the mappingto a BAMS nomenclature was performed by the collator

284 __________________________________________________________________________Bota and Swanson


insertion of records in the ldquoMoleculesrdquo modulewhereas the other metadata classes are optionalThe third step of BAMSrsquos curation policy con-sists of records and metadata checking per-formed by the referencersquos author Onceinformation collated from a reference is com-pletely inserted in BAMS curators contactauthors to verify the accuracy consistency andcompleteness of inserted data and metadata Alldata and metadata inserted in BAMS andchecked by authors of collated references istagged and displayed accordingly in the back-end database and the user interface respectively

BAMSrsquos ldquoMoleculesrdquo module also includes aldquoPersonal Accountrdquo part that allows registeredusers to insert data from collated references orunpublished experimental data The databasestructure of this part is similar to the public partof ldquoMoleculesrdquo and the insertion process fol-lows the knowledge representation schemashown in Fig 1 The conceptual schema of dataand metadata insertion and update in theldquoPersonal Accountrdquo section is shown in Fig 4

Results

Searching for Molecules in BAMS

Users can search for information related tomolecules compare the presence of moleculesin different CNS regions search for informa-tion by type of manipulation and constructreports from multiple experiments The searchfor information related to molecules identi-fied in CNS parts is performed by using a formthat displays all molecules inserted in BAMSarranged according to the classificationschema The result of this search is a list ofCNS parts where the searched molecule wasidentified and associated experimental con-ditions Details about the presence of a mole-cule in a CNS region can be accessed byclicking on the link associated with the exper-imental condition (Fig 5) As shown in Fig 5users can access the metadata associated witha record An example of metadata associatedwith retrieved records and accessible by usersis shown in Fig 6

Fig 3The conceptual schema of the curation policy implemented in BAMS



Users can also view all molecules identifiedin a brain region when they search for CNSregions The result of a search for CNS regionsin BAMS is a page that summarizes data col-lated by researchers or inferred by the systemand associated with the part of interest Adetailed description of BAMS web interfacesis provided in Bota et al (2005) If a CNS regionis associated in BAMS with several moleculesthey are returned in the page summarizing theregion as shown in Fig 7 Users can access fur-ther information about each retrieved mole-cule by clicking on the associated links (Fig 8)

Records associated with the molecule ofinterest are grouped in four categories (mole-cule presence in normal and manipulated phys-iological state and coexpression of othermolecules in normal and manipulated state seeFig 8 by experimental condition and by coex-pression of other molecules in the same CNSregion) The system first retrieves referencescorresponding to each category Every refer-ence is associated with a ldquoDatardquo link thatretrieves corresponding experimental recordsIf at least two experiments from different

references are retrieved in the same categoryand have metadata associated with them a newlink called ldquoMethods comparisonrdquo becomesavailable Users can therefore compare meth-ods used in all experiments associated withmetadata retrieved in one of the four categoriesAn example of methods comparison is shownin Fig 9

The methods comparison across referencesand experiments for the same CNS region mol-ecule and experimental condition firstretrieves metadata associated with eachretrieved reference data types mappingapproach and data presentation This allowsusers to compare rapidly and evaluate exper-iments and references in terms of data richnessand mapping consistency The second comparison is performed across experimentmetadata And the third comparison allowsusers to compare metadata associated withretrieved experimental series In short thisBAMS user interface functionality can beused by users to compare the accuracy andreliability of molecular data by comparingassociated metadata

Fig 4The conceptual schema of the data entry part of Moleculesrsquo ldquoPersonal accountrdquo part

286 __________________________________________________________________________Bota and Swanson


If a molecule identified in a CNS region isassociated in BAMS with time-dependent datausers can access this information by clickingon a link called ldquoTime dependencerdquo (Fig 8)The time-dependent molecular data aregrouped by experiment reference and exper-imental state of the animals The statistical sig-nificance associated with the quantities beingcompared is also retrieved and is graphically

coded An example of time-dependent dataretrieval is shown in Fig 10

BAMSrsquos ldquoMoleculesrdquo module includes aninference engine for reconstructing thechemoarchitectonic profile of a CNS regionfrom the molecular data associated with itssubstructure This engine can be accessed fromthe page describing CNS parts in BAMSshown in Fig 6 It is similar to the projections

Fig 5The result of search of CNS regions where the neurotransmitter corticotropin-releasing hormone wasidentified Users can view details of records associated with each retrieved CNS region including qualitativedensity cell counts statistical measurements spatial characteristics of cells expressing the molecule and asso-ciated annotations by clicking on links associated with experimental conditions For reports associated withmanipulated state users can also view type of manipulation injected chemicals and details about experimentalprocedure Users may also access metadata associated with retrieved records Records verified by originalauthors of collated references will also have a check mark associated with them



profile inference engine described in Bota et al(2005) and it displays gene expression data asa function of experimental conditions (normaland manipulated states) An example chemo-architectonic profile reconstruction is shownin Fig 11

BAMSrsquos ldquoMoleculesrdquo module also allowsconstruction of reports from references andexperiments of interest Users can chooseexperiments individually and reconstruct theassociated data in matrix format The outputof such customized reports is similar to thematrix shown in Fig 11

Comparing Molecules in BAMS

The ldquoMoleculesrdquo module user interfaceincludes two options for comparing the pres-ence of various molecules in particular CNSregions from inserted data The first optioncalled ldquoComparison of molecules existence inCNS regionsrdquo processes the query type ldquoWhatare the CNS parts where all the molecules ofinterest have been identifiedrdquo This enginereturns a list of CNS parts where all the cho-sen molecules have been identified and theassociated physiological states This engine alsoreturns records of coexpression data for any

Fig 6 Users can access metadata associated with retrieved experimental data and with the correspondingexperiment and experimental series

288 __________________________________________________________________________Bota and Swanson


pair of molecules in the chosen set An exam-ple of the reconstruction of the molecular com-position of CNS parts is shown in Fig 12

The second option called ldquoSearch by type ofmanipulationrdquo processes the query typeldquoShow identified molecules and CNS parts thatare associated with a specific molecule or struc-tural manipulationrdquo This engine returns thelist of CNS parts and molecules associated withthe searched type of manipulation as well asbasal state data for each retrieved molecule Ifany retrieved experimental data is associatedwith statistical information it also will be dis-played in graphical format An example of the

online retrieval of chemoarchitectonic dataassociated with systemic injection of polyeth-ylene glycol is shown in Fig 13

The BAMS Workspace

The functionality of BAMSrsquos web interfacewas augmented with two additional featuresregistered users may now add comments todata reports and they can also save their activ-ity in a personal workspace Registered usersare allowed to attach comments to brainregion projection and molecule reportsinserted in the public part of BAMS Thesecomments may be accessed and viewed at any

Fig 7 The result of a CNS parts search in BAMS If a CNS region is associated in BAMS with moleculesthe latter are listed under the category ldquoChemoarchitecturerdquo and divided according to Moleculersquos classification schema



time by the registered user who inserted themThe BAMS workspace now becomes the placewhere registered users can save and viewreports of interest concerning CNS regionstheir input and output axonal connections cus-tomized connection matrices (Bota et al 2005)molecules and groups of gene expression

pattern experiments The number of reportsand matrices that can be used by registeredusers is unlimited An example of the types ofinformation that can be saved in the BAMSworkspace is shown in Fig 14

The BAMS workspace combined with theoption of adding comments to data reports

Fig 8 Users can view detailed reports about the presence of a molecule in a particular CNS regionExperimental data is grouped in four categories which depend on the physiological state of the animal andcoexpression data Records verified by original authors are indicated with a green checkmark

290 __________________________________________________________________________Bota and Swanson


brings a new level of interactivity to the sys-tem and allows registered users to annotatereports according to their expertise

Inserting Data in BAMSrsquosldquoMoleculesrdquo Module

The ldquoMoleculesrdquo modulersquos ldquoPersonalaccountrdquo functionality allows registered usersto insert molecular data from published refer-ences or unpublished experiments updateinserted records view inserted data combineunpublished experiments with published ref-erences and create reports from references andexperiments of interest Registered BAMS users

may thus be collators andor authors andorcurators Insertion of molecular data in aldquoPersonal accountrdquo can be performed with abroad range of complexity from very simplereports of presence or absence in a CNS regionto expression in different neuron types to verycomplex reports when all metadata are speci-fied The structure of ldquoMoleculesrdquo modulersquosldquoPersonal accountrdquo (see Fig 3 for its knowledgerepresentation) also allows updating previ-ously inserted information when adding newrecords and the addition of new fields Anexample of data entry in a ldquoPersonal Accountrdquois shown in Fig 15

Fig 9 Users may compare metadata associated with experimental data and experiments for the presence ofa molecule in a CNS region of interest



Transfer of inserted data and metadata fromthe ldquoPersonal accountrdquo to the public part ofBAMS is performed according to theldquoMoleculesrdquo Module curation policy On com-pleting the insertion of a reference the colla-tor verifies its accuracy and completeness andthen contacts the referencersquos authors if possi-ble After the curator checks the reference andits associated records and metadata they aretransferred to the public part of BAMS with-out being erased from the collatorrsquos ldquoPersonalaccountrdquo Once the data is transferred to thepublic part of BAMS collators are notified ofthis action

Collators may also combine unpublishedexperiments with data from published refer-ences In this case curators do not check theconsistency and completeness of recordsinserted by authors The ldquoPersonal accountrdquofunctionality also allows creating reports fromexperiments of interest to an individual colla-tor This option is similar to the one imple-mented in BAMSrsquos public part

Inserted Data

The BAMS ldquoMoleculesrdquo module containsmore than 3000 reports about the presence of 18molecules in the rat CNS Because all molecular

Fig 10 The display of time-dependent chemoarchitectonic data is grouped by experiment reference andphysiological state of the subjectUsers can view qualitative assessmentsquantitative measurementsand anno-tations associated with each retrieved record as well as the statistical significance of each experimental pointThe retrieved statistical significance is graphically coded and associated with quantities compared within oracross experiments

292 __________________________________________________________________________Bota and Swanson


data inserted so far in BAMS originated fromour laboratory or from research groups thatuse the Swanson-1998 rat CNS nomenclaturefor mapping (Swanson 1998) they were reg-istered to this nomenclature in an internallyconsistent way

Overall BAMS contains 11 comprehensiveCNS nomenclatures from five species humanmacaque (Macaca fascicularis) cat rat andmouse approx 40000 reports of neu-roanatomical projections as collated from theliterature since 1962 and related generally to

the visual and limbic systems of the rat 4000gene expression reports and data referring to175 neuronal cell types identified in 50 rat brainregions that are defined in the Swanson-1998nomenclature (Swanson 1998) This data wasinserted in BAMS by five collators and twocurators

Conclusions

This article describes the structure and majorfeatures of BAMSrsquos ldquoMoleculesrdquo module whichhas been designed for the online handling of

Fig 11The result of a BAMS gene expression pattern reconstruction dealing with the various subdivisions ofthe rat paraventricular nucleus of the hypothalamusThe reconstructed chemoarchitecture profile includespresence or absence of molecules in both normal and manipulated experimental conditions Users can viewdetails of the associated reports by clicking on the symbols in each cell A similar output is generated forreports created by users from experiments of interest



molecular and gene expression data associatedwith different CNS regions or cell types andcollated from the literature or from a collatorrsquospersonal unpublished results

The first major new contribution is theldquoMoleculesrdquo module knowledge representa-tion which completely separates data frommetadata in BAMS and is detailed enough toallow association of different types of metadata

with a collated reference At the same time theknowledge representation now implementedin ldquoMoleculesrdquo is general enough to allow exten-sions with other gene expression techniquesand to other types of neuroscientific experi-mental data Thus the classes ldquoExperimentmetadatardquo ldquoData presentation metadatardquo andldquoMapping metadatardquo can be applied to a widerange of experiments on different parts of the

Fig 12 Users may compare the existence of particular molecules in different CNS regionsThe result is a listof regions that are associated in BAMS with reports of all the searched molecules (corticotropin-releasinghormone [CRH] oxytocin [Oxy] and vasopressin [VAS]) in either basal physiological state or manipulatedstate Coexpression data is also returned for all pairs of molecules in the set of interest Users can accessreports of molecule presence in the retrieved CNS regions

294 __________________________________________________________________________Bota and Swanson


vertebrate CNS The knowledge representationrsquoshierarchical structure also allows extension ofmetadata that describe records about moleculesinserted in BAMS and association of new meta-data classes for the description of new experi-mental techniques

The curation policy now implemented inBAMS allows rigorous collation of data frompublished references and full documentationof actions performed by collators The prob-lems of data accuracy and consistency which

are very important in populating neuroinfor-matics databases are addressed by the set ofcuration rules described here This set requiresthe rigorous mapping of molecular data ontoCNS nomenclatures in BAMS and includesvalidation of inserted data by original authors

However this set of curation rules does notsolve the problem of contradictory resultswhich are common in the neuroscience litera-ture Automatic evaluation of molecular exper-imental data is not yet possible unlike the case

Fig 13 Users can retrieve chemoarchitectonic data associated with a chemical or structural manipulation andcompare the presence of molecules in basal and manipulated states



for neuroanatomical axonal connection reports(Bota and Arbib 2004) because experimentaltechniques are quite diverse and complicatedand results also depend on experimental con-ditions like time of day and behavioral stateMoreover techniques and methods used forgene expression mapping continue to evolveThus potential automatic reliability evalua-tion will have to change accordingly

Nevertheless accuracy evaluation for molec-ular experimental data can be performed inBAMS For this users compare metadata asso-ciated with experiments according to theirexpertise We present here a relatively compre-hensive and extendable set of metadata thatdescribe the most widely used experimental tech-niques a range of experiment details and therichness of data collated from the literaturemdashall

of which can be accessed by users for the com-parison and evaluation of gene expression andother molecular data Moreover users can gen-erate reports from data recorded in BAMS bychoosing only gene expression experiments thatare accuratemdashaccording to their expertiseFinally registered users may annotate data withcomments and save the results of this activityin their personal workspace for future use BAMSto our knowledge is the first online neuroinfor-matics system that allows users to interact withthe system by annotating inserted informationand storing the results of their searches

The structure of BAMSrsquos ldquoMoleculesrdquo mod-ule allows insertion of complex reports aboutmolecules identified in CNS parts or expressedin different cell types It also allows constructionof a web interface enabling users to search for

Fig 14The BAMS workspace allows registered users to store and view reports of interest concerning CNSparts and cell types input and output axonal connections and a wide variety of moleculesmdashas well as com-plex axonal connection matrices and groups of gene expression patterns

296 __________________________________________________________________________Bota and Swanson


information related to the chemical makeupof CNS regions of interest run complex queriesthat reconstruct gene expression patterns indifferent CNS regions and compare the exis-tence and levels of molecules under specifiedphysiological conditions

This is part of an expanding effort in the emerg-ing field of neuroinformatics Several researchgroups have developed on-line knowledge man-agement systems that handle data at differentlevels of nervous system organization brain

region nomenclatures (Bowden and Martin1997 Burns 2001 Stephan et al 2001 Bota andArbib 2004) neuroanatomical projections(Burns 2001 Stephan et al 2001 Bota and Arbib2004) and cytology (Marenco et al 1999 Botaand Arbib 2004) Neuroinformatics systems thatshare similar features with BAMS includeNeuroNames (httpbraininfo rprcwashing-tonedu Bowden and Dubach 2003) CoCoMac(httpcocomacorg Koumltter 2004) and theNeuronDB database (http senselabmed

Fig 15The BAMS ldquoPersonal accountrdquo allows registered users to insert update and view molecular data andmetadata from collated references (or the results of unpublished experiments) to group unpublished exper-iments with new published references and to create reports about experiments of interest



yaleedusenselabNeuronDB Marenco et al1999) However none of these systems includea representation of molecular data identified invarious CNS parts and cell types BAMS appearsto be the first on-line neuroinformatics systemthat manipulates neuroscience data across fourlevels of CNS organization molecules cell typesbrain regions and networks of brain regions

The present status of BAMS is ldquoin progressrdquobecause we continue to add new data and extendits functionality High priority future develop-ments of the ldquoMoleculesrdquo module include exten-sion of the present molecule classificationschema molecule representation at the level ofsynapses (and thus terminal fields) and infer-ence engines for automatic comparison of molec-ular data according to the age sex and othermetadata of experimental animals Finallyimplementation of webservices protocols forproviding data and process queries submittedin software applications developed by otherresearch laboratories (MacKenzie-Graham et al2004) is a top priority BAMS is already a dataprovider for a series of neuroinformatics sys-tems through URL links (NeuroNames) mid-dlelayer applications (LONI) and backendMySQL connections (NeuronDB LONIWebQTL) Our goal is to continue expandingthe list of neuroinformatics and bioinformaticssystems to which BAMS acts as data provideror client including gene expression databasesdeveloped by the Allen Institute for BrainScience (httpwwwbrainatlasorg) andGensat (httpwwwgensatorg)

Acknowledgments

This work was supported by NIH GrantsMH61223 NS16668 and NS050792-01 Weespecially thank Drs Alan Watts and ArshadKhan for their suggestions and feedback

ReferencesBota M Dong H W and Swanson LW (2003)

From gene networks to brain networks NatureNeurosci 6 795ndash799

Bota M Dong H W and Swanson LW (2005)The Brain Architecture Management SystemNeuroinformatics 3 15ndash48

Bota M and Arbib M A (2004) IntegratingDatabases and Expert Systems for the Analysisof Brain Structures Connections Similarities andHomologies Neuroinformatics 2 19ndash58

Bowden D M and Martin R F (1997) A digitalRosetta stone for primate brain terminology InBloom F E Bjorklund A and Hokfelt T (eds)Handbook of Chemical Neuroanatomy vol 13The Primate Nervous System part I AcademicPress San Diego pp 1ndash37

Bowden D M and Dubach F M (2003)NeuroNames 2002 Neuroinformatics 2 63ndash83

Burry R W (2000) Specificity controls for immuno-cytochemical methods J Histochem Cytochem48 163ndash165

Burns G A P C (2001) Knowledge mechanics andthe NeuroScholar project a new approach toneuroscientific theory In Arbib M A andGrethe J (eds) Computing the Brain A Guideto Neuroinformatics Academic Press San Diegopp 319ndash336

Burt D R (1985) Criteria for receptor identifica-tion In Yamamura et al (eds) Neurotransmitterreceptor binding 2nd ed Raven Press NewYork pp 41ndash60

Koumltter R (2004) Online retrieval processing andvisualization of primate connectivity data from theCoCoMac Database Neuroinformatics 2 127ndash144

Kuhar M J (1985) Receptor localization with themicroscope in Yamamura et al (eds)Neurotransmitter receptor binding 2nd edRaven Press New York pp 41ndash60

Kuhar M J De Souza E B and Unnerstall J R(1986) Neurotransmitter receptor mapping byautoradiography and other methods Annu RevNeurosci 9 27ndash59

Marenco L Nadkarni P Skoufos E Shepherd Gand Miller P (1999) Neuronal database integra-tion the Senselab EAV data model Proceedingsof AMIA Symposium pp 102ndash106

MacKenzie-Graham A Lee E F Dinov I D et al(2004) A multimodal multidimensional atlas ofthe C57BL6J mouse brain J Anat 204 93ndash102

MacKenzie-Graham A Jones E S Shattuck D WDinov I D Bota M and Toga A W (2003) Theinformatics of a C57BL6J mouse brain atlasNeuroinformatics 1 397ndash410

Saper C B (2005) An open letter to our readers onthe use of antibodies J Comp Neurol 493477ndash478

298 __________________________________________________________________________Bota and Swanson


Saper C B and Sawchenko P E (2003) Magic pep-tides magic antibodies guidelines for appropriatecontrols for immunohistochemistry J CompNeurol 465 161ndash163

Stephan K E Kamper L Bozkurt A Burns G AP C Young M P and Koumltter R (2001) Advanceddatabase methodology for the Collation ofConnectivity data on the Macaque brain(CoCoMac) Philos Trans R Soc London B BiolSci 1159ndash1186

Swanson L W (1998) Brain Maps Structure of theRat Brain Second Edition Elsevier San Diego

Swanson L W (2000a) What is the brain TrendsNeurosci 23 519ndash527

Swanson L W (2000b) Ahistory of neuroanatomicalmapping In Brain Mapping The ApplicationsToga A W and Mazziotta J C (eds) AcademicPress San Diego pp 77ndash109

Swanson L W and Simmons D M (1989) Differentialsteroid hormone and neural influences on

peptide mRNA levels in CRH cells of theparaventricular nucleus a hybridizationhistochemical study in the rat J Comp Neurol 285413ndash435

Wada E Wada K Boulter J et al (1989)Distribution of alpha 2 alpha 3 alpha 4 and beta2 neuronal nicotinic receptor subunit mRNAs inthe central nervous system a hybridization his-tochemical study in the rat J Comp Neurol 284314ndash335

Watts A G and Sanchez-Watts G (2002)Interactions between heterotypic stressors andcorticosterone reveal integrative mechanisms forcontrolling corticotropin-releasing hormonegene expression in the rat paraventricularnucleus J Neurosci 22 6282ndash6289

Watts A G and Swanson L W (1989) Combinationof in situ hybridization with histochemistry andretrograde tract-tracing Methods in neurosciencesvol I Academic Press San Diego pp 127ndash136

276 __________________________________________________________________________Bota and Swanson


Introduction

The vast amount of neuroscience-related tex-tual and graphical informationmdashwhich actu-ally extends back centuriesmdashis one importantfactor that makes searching and interpretationa difficult task In addition the interpretationand integration of data are derived from dif-ferent levels of central nervous system (CNS)organization Any CNS region can be describedwith respect to multiple levels of organizationfrom patterns of gene expression under spe-cific experimental conditions to structuralchemical and physiological characteristics ofits constituent neuronal and glial cell types tospecific roles in functional networks of regionsin a particular species (Swanson 2000ab Botaet al 2003)

Database systems are complementary to theneuroinformatics visualization tools and appli-cations developed by individual researchgroups and are necessary for data storage andhandling The display of multiple modalitiesmdashincluding structural and functional data andgene expression patterns (levels of expressedmolecules and in situ hybridization data)mdashonthree-dimensional brain reconstructions ortwo-dimensional atlas reconstructions ofbrains from various species requires the designand development of complex backend data-bases as information repositories

To address these problems we are develop-ing the on-line knowledge management systemBrain Architecture Management System(BAMS) for systematizing organizing and pro-cessing neuroscience information relevant todifferent levels of CNS organization The struc-ture of BAMS allows insertion of data about CNSregions cell types neural (axonal) pathwaysand molecules collated from the literature orrecorded by users BAMS includes inferenceengines that establish qualitative topologicalrelations between CNS regions reconstruct pro-jection patterns from CNS regions of interestand construct possible networks of CNS regions

from connectivity data resident in the system(Bota et al 2005)

BAMS is also a data provider for neuroinfor-matics systems developed by other groups Forexample simple data transfer protocols havebeen implemented with several web-based andstand-alone applications created by theLaboratory of NeuroImaging (LONIhttpwwwloniuclaedu) at UCLA that provide hierarchically organized neuroanatomicalnomenclatures and definitions of CNS parts(MacKenzie-Graham et al 2003 2004)

Here we describe the structure of a newBAMS module ldquoMoleculesrdquo that allows inser-tion of molecular (chemical) data pertaining tovarious CNS regions and cell types We alsodescribe here the user interfaces designed toallow searching for molecules in BAMS whosedatabase structure and inference engines aredescribed elsewhere (Bota et al 2005)

Methods

BAMS is hosted on a Dell 8200 desktop com-puter with a Pentium III processor runningunder the Windows XP operating system withIIS 51 The backend relational database of BAMSwas created in MySQL and PHP was used toconstruct web interfaces and inference enginesThe architecture of BAMS is constructed on threelevels a set of tables constructed in MySQL thatstore data collated from the literature (or insertedby neuroanatomists) an intermediate levelencoded in PHP that includes queries and algo-rithms for processing data populating BAMSand an output level that is mainly in an HTMLtabular format and graphics

ldquoMoleculesrdquo Module Database Structure

The BAMSrsquos ldquoMoleculesrdquo module consistsof two parts a public part and a ldquoPersonalaccountrdquo part that was designed to store dataentered by collators and restricted to their useThe knowledge representation schema ofBAMSrsquos ldquoMoleculesrdquo module (Fig 1) follows










278 __________________________________________________________________________Bota and Swanson
























280 __________________________________________________________________________Bota and Swanson



Metadata




















282 __________________________________________________________________________Bota and Swanson





























284 __________________________________________________________________________Bota and Swanson




Results











286 __________________________________________________________________________Bota and Swanson













288 __________________________________________________________________________Bota and Swanson





The BAMS Workspace









290 __________________________________________________________________________Bota and Swanson











Inserted Data



292 __________________________________________________________________________Bota and Swanson





Conclusions









294 __________________________________________________________________________Bota and Swanson














296 __________________________________________________________________________Bota and Swanson










Acknowledgments


















298 __________________________________________________________________________Bota and Swanson





















278 __________________________________________________________________________Bota and Swanson
























280 __________________________________________________________________________Bota and Swanson



Metadata




















282 __________________________________________________________________________Bota and Swanson





























284 __________________________________________________________________________Bota and Swanson




Results











286 __________________________________________________________________________Bota and Swanson













288 __________________________________________________________________________Bota and Swanson





The BAMS Workspace









290 __________________________________________________________________________Bota and Swanson











Inserted Data



292 __________________________________________________________________________Bota and Swanson





Conclusions









294 __________________________________________________________________________Bota and Swanson














296 __________________________________________________________________________Bota and Swanson










Acknowledgments


















298 __________________________________________________________________________Bota and Swanson































280 __________________________________________________________________________Bota and Swanson



Metadata




















282 __________________________________________________________________________Bota and Swanson





























284 __________________________________________________________________________Bota and Swanson




Results











286 __________________________________________________________________________Bota and Swanson













288 __________________________________________________________________________Bota and Swanson





The BAMS Workspace









290 __________________________________________________________________________Bota and Swanson











Inserted Data



292 __________________________________________________________________________Bota and Swanson





Conclusions









294 __________________________________________________________________________Bota and Swanson














296 __________________________________________________________________________Bota and Swanson










Acknowledgments


















298 __________________________________________________________________________Bota and Swanson
























282 __________________________________________________________________________Bota and Swanson





























284 __________________________________________________________________________Bota and Swanson




Results











286 __________________________________________________________________________Bota and Swanson













288 __________________________________________________________________________Bota and Swanson





The BAMS Workspace









290 __________________________________________________________________________Bota and Swanson











Inserted Data



292 __________________________________________________________________________Bota and Swanson





Conclusions









294 __________________________________________________________________________Bota and Swanson














296 __________________________________________________________________________Bota and Swanson










Acknowledgments


















298 __________________________________________________________________________Bota and Swanson





























284 __________________________________________________________________________Bota and Swanson




Results











286 __________________________________________________________________________Bota and Swanson













288 __________________________________________________________________________Bota and Swanson





The BAMS Workspace









290 __________________________________________________________________________Bota and Swanson











Inserted Data



292 __________________________________________________________________________Bota and Swanson





Conclusions









294 __________________________________________________________________________Bota and Swanson














296 __________________________________________________________________________Bota and Swanson










Acknowledgments


















298 __________________________________________________________________________Bota and Swanson



















286 __________________________________________________________________________Bota and Swanson













288 __________________________________________________________________________Bota and Swanson





The BAMS Workspace









290 __________________________________________________________________________Bota and Swanson











Inserted Data



292 __________________________________________________________________________Bota and Swanson





Conclusions









294 __________________________________________________________________________Bota and Swanson














296 __________________________________________________________________________Bota and Swanson










Acknowledgments


















298 __________________________________________________________________________Bota and Swanson



















288 __________________________________________________________________________Bota and Swanson





The BAMS Workspace









290 __________________________________________________________________________Bota and Swanson











Inserted Data



292 __________________________________________________________________________Bota and Swanson





Conclusions









294 __________________________________________________________________________Bota and Swanson














296 __________________________________________________________________________Bota and Swanson










Acknowledgments


















298 __________________________________________________________________________Bota and Swanson


















290 __________________________________________________________________________Bota and Swanson











Inserted Data



292 __________________________________________________________________________Bota and Swanson





Conclusions









294 __________________________________________________________________________Bota and Swanson














296 __________________________________________________________________________Bota and Swanson










Acknowledgments


















298 __________________________________________________________________________Bota and Swanson
















Inserted Data



292 __________________________________________________________________________Bota and Swanson





Conclusions









294 __________________________________________________________________________Bota and Swanson














296 __________________________________________________________________________Bota and Swanson










Acknowledgments


















298 __________________________________________________________________________Bota and Swanson












292 __________________________________________________________________________Bota and Swanson





Conclusions









294 __________________________________________________________________________Bota and Swanson














296 __________________________________________________________________________Bota and Swanson










Acknowledgments


















298 __________________________________________________________________________Bota and Swanson


















294 __________________________________________________________________________Bota and Swanson














296 __________________________________________________________________________Bota and Swanson










Acknowledgments


















298 __________________________________________________________________________Bota and Swanson



















296 __________________________________________________________________________Bota and Swanson










Acknowledgments


















298 __________________________________________________________________________Bota and Swanson
















Acknowledgments


















298 __________________________________________________________________________Bota and Swanson












298 __________________________________________________________________________Bota and Swanson












Original Article · 2016. 1. 31. · Chemoarchitecture Chemoarchitecture reports as collated from...

Documents

Transcript of Original Article · 2016. 1. 31. · Chemoarchitecture Chemoarchitecture reports as collated from...