This document is the Accepted Manuscript version of a Published Work that appeared in final form in Journal of Chemical Article
Information and Modeling, copyright © American Chemical Society after peer review and technical editing by the
pubs.acs.org/jcim
publisher. To access the final edited and published work see http://dx.doi.org/10.1021/ci500593j

Managing the Computational Chemistry Big Data Problem: The
ioChem-BD Platform
M. Á lvarez-Moreno,*,†,‡ C. de Graaf,‡,∥ N. López,† F. Maseras,†,§ J. M. Poblet,‡ and C. Bo*,†,‡
†

Institute of Chemical Research of Catalonia, ICIQ, Av. Països Catalans 16, 43007 Tarragona, Catalonia, Spain
Department of Physical and Inorganic Chemistry, Universitat Rovira i Virgili, C/Marcel·lí Domingo s/n, 43007 Tarragona,
Catalonia, Spain
§
Department of Chemistry, Universitat Autònoma de Barcelona, 08193 Bellaterra, Catalonia, Spain
∥
Catalan Institution for Research and Advanced Studies, ICREA, Passeig Lluis Companys 23, 08010 Barcelona, Catalonia, Spain
Downloaded via UNIV ROVIRA I VIRGILI on April 17, 2019 at 16:14:34 (UTC).
See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

‡

ABSTRACT: We present the ioChem-BD platform (www.
iochem-bd.org) as a multiheaded tool aimed to manage large
volumes of quantum chemistry results from a diverse group of
already common simulation packages. The platform has an
extensible structure. The key modules managing the main tasks
are to (i) upload of output files from common computational
chemistry packages, (ii) extract meaningful data from the
results, and (iii) generate output summaries in user-friendly
formats. A heavy use of the Chemical Mark-up Language
(CML) is made in the intermediate files used by ioChem-BD.
From them and using XSL techniques, we manipulate and
transform such chemical data sets to fulfill researchers’ needs in
the form of HTML5 reports, supporting information, and other
research media.

1. INTRODUCTION
Intensive high performance computing is one of the pillars to
accelerate materials discovery and development in many fields
of science and engineering, most prominently chemistry,
physics, and related areas. The volume of information
generated daily, coming from the results of scientific
calculations, is increasing exponentially. For instance, in our
lab, scientists (a group of about 10) generate 1.3TB weekly. Its
conservation on physical media is favored right now by the
cheap price of storage per bit of information as well as an
increase in the available telecommunications infrastructure
bandwidth. This fact makes the public and private centers
generate and store more and more terabytes of information. In
our particular case, this information corresponds to the
outcome of calculations. At present, storage of computational
simulations has not been identified as a bottleneck in most data
and supercomputing centers. However, the main global players
in the Internet business have already introduced the “Big
Data”1 concept to start looking for solutions to maintain all
physical data storage systems sustainable and provide
convenient access. Solutions based on what is called “the
cloud” along with concepts derived from the “social networks”
are leading to a reformulation of how and in what physical
space the information shall be stored. As a result, there is a
growing demand for tools to order the storage, allow analysis,
and simplify the presentation of significantly large volumes of
growing data in an amenable, transparent, and reliable manner.2
© 2014 American Chemical Society

Only a very small percentage of the information currently
stored in data centers is hierarchically indexed.3 This simply
means that the information available is impossible to process;
the information bits are hardly usable by any person other than
the creator himself. Tim Berners-Lee, one of the original
creators of the World Wide Web, identified the need to
transform numerical data into “raw data”.4 This raw data is the
desirable state where information is meaningful because it is
enriched with labels that contextualizes it, the labels being
“metadata”. Once contextualized, searching the ocean of
information is more efficient. Moreover, the search process
becomes a process of knowledge creation, as it is possible to
establish new connections, in what is already called “linked
data”.5
The application of these concepts to the field of computational chemistry is hindered by the heterogeneity of the
packages used in the atomistic simulations of molecules and
materials. As a result, the outcome of atomistic simulations
addressing chemical or physical problems and based on the
application of the Schrödinger equation are presented in a
disperse manner, with sparse data showing only some of the
key aspects as geometries, energies, and chemical and/or
physical properties. The wave function or density data does not
belong to this set due to the following reasons: (i) the size of
these files is excessive for our purposes; (ii) the speed of the
Received: September 30, 2014
Published: December 3, 2014
95

dx.doi.org/10.1021/ci500593j | J. Chem. Inf. Model. 2015, 55, 95−103

Journal of Chemical Information and Modeling

Article

Scheme 1. ioChem-BD System Overviewa

a

Web nature of the CREATE and BROWSE modules is highlighted. Both modules share functionalities like searching and browsing chemical
datasets, exporting to third party formats, and publishing results, but without losing sight of the original private−public sense of each module.

2. BASE TECHNOLOGIES
The keystone in the definition of the project is that it employs
high reliability software technologies widely used in the
Internet world that are extended to cover our particular area
of interest. We chose eXtensible Markup Language (XML)12 as
the container element of all information for its reliability,
format neutrality, and ease of validation (using XSD verification
tools).13 To be more specific, we chose Chemical Markup
Language (CML) implementation because it contains all
semantics necessary to describe most chemical entities.14
With calculations in CML format, querying its content for
specific information is extremely easy and efficient by using
XPATH queries.15 Working with XML provides a wide range of
conversion operations from CML files into any other existing
or future format using eXtensible Stylesheet Language
Transformations (XSLT).16
As access to information is becoming more universal, the
data in the system is reachable through the Internet by any
digital device on the market, following the latest existing Web
standards in communication.12,17 Users have at their disposal
the latest search,18,19 display,20 and data-labeling tools,14 and
also, there exists communication channels enabled to propose
new features.
In terms of data storage, information is distributed among
the content generators, creating a mesh topology in which the
service is always available and accessible on the network. Being
a cloud system, it has the necessary standards in data

computers, once the optimized geometries are available, allows
the recreation of this massive data. The ultimate consequence is
that the data published in the scientific journals of chemistry,
physics, nanoscience, biochemistry and related areas are not
homogeneous; they are often incomplete, are hardly consulted
in bulk, and are rarely reused. The high degree of diversity in
the data formats requires the definition of standards.6 There
have been technological initiatives7 like the Quixote project that
implement solutions on multiple aspects of this problem like
data format unity and data management.8 Parallel initiatives to
ours are under development,9 for instance, the AiidA software,
that according to the available documentation focuses on
elaborated workflows to carry out complex calculations in an
automated manner. Other purpose-dedicated databases have
been generated by the groups of MIT, Berkeley,10 and
Standford.11
In this paper, we present an alternative platform, ioChemBD, encompassing a variety of aspects in the definition of
standards for treatment, hierarchical storage, and retrieval of
data. Our platform automates the extraction of relevant data
and its conversion into fully tagged information in a distributed
database. It provides tools for the researcher to validate, enrich,
publish, and share information, and tools in the cloud to access
it and view it.
96

dx.doi.org/10.1021/ci500593j | J. Chem. Inf. Model. 2015, 55, 95−103

Journal of Chemical Information and Modeling

Article

definition21−23 in order to connect to other digital repositories
and external Web services to build up a network of
interconnected semantic data to provide the most sense to
the user experience.
Finally, to enhance industrial implementation, the platform
allows secure24,25 and reliable channels for the communication
and collaboration between users, groups, and/or centers but
with the highest privacy standards for third partners. All
information has configurable levels of access and licensing,
allowing for adaptation to the specific legal needs of each entity.
Architecture. The ioChem-BD platform is composed of
two main modules that work independently, labeled CREATE
and BROWSE. Both of them are executed as Java Web services.
The CREATE module is designed to extract the information
from the output files generated by the computational chemistry
packages and store it in an organized way. It currently manages
output files from the programs Gaussian,26 ADF,27 and VASP.28
The list of accessible codes should be expanded in the midterm
future to codes such as SIESTA, 29 TURBOMOLE, 30
MOLCAS,31 and ORCA.32 The CREATE module shall be
used by the scientists that execute the simulations (creator).
The BROWSE module is designed as a tool to explore and
use the data contained in the database, and it resides in the
cloud. The BROWSE module has a much broader scope and is
useful to researchers interested in accessing the computational
Big Data.
The combination of both modules gives the scientific
community storage and access to all the Chemistry Big Data
in a way schematically shown in Scheme 1.
CREATE Module. The CREATE module contains two
different subunits that allow: (i) the extraction and structuring
of the relevant output data and (ii) the publication of such data
and derivates into BROWSE.
CREATE Module: Data Extraction from Output Files. The
system initially works with input and output files from the
computational chemistry packages described above. In Gaussian
and ADF, the results are stored in a single file that needs to be
extracted. However, for other computational codes, the
CREATE module needs a group of files. This is the particular
case for VASP calculations, where relevant data are split into
multiple files from inputs like POTCAR or output files
OUTCAR (summary), CONTCAR (geometries), XDATCAR
(trajectory), and vasprun.xml. The ioChem-BD platform
outputs a single CML file from these sources. The upload of
these files to the CREATE module is done inside a layered
process where we simultaneously parse and tag all relevant data;
we infer its metadata and capture the molecular geometry. A
scriptable shell upload utility is used to do file conversion and
upload calculations straight from HPC clusters to the CREATE
module; there is also an alternative mechanism to upload
content from the user Web browser. A detection algorithm
decides which format extraction templates to use based on the
calculation file content. Once the file format is elucidated, the
appropriate templates for such formats are selected, and the
first conversion is performed from plain text to CML using a
modified version of the JUMBOConverters library.33 This is
followed by a second conversion to reorder CML tags so they
comply with the CompChem convention.34 This process can
be used on individual or multiple output files as is depicted in
Figures 1 and 2.
As an additional feature, the module is able to attach directly
to other supporting files such as calculation input files, graphics,
and text, and all needed associated gray literature. These

Figure 1. Conversion workflow for individual output files. This twostep process is the default behavior for file conversion inside the
JUMBOConverters library, from output files into CML elements, and
then to compliant CML CompChem.

Figure 2. Conversion workflow for multiple output files. Our
customized JUMBOConverters library accepts multiple output files
for its unification into a single CML file. Uploaded files can be a
mixture of plain text and XML files.

97

dx.doi.org/10.1021/ci500593j | J. Chem. Inf. Model. 2015, 55, 95−103

Journal of Chemical Information and Modeling

Article

Figure 3. CREATE main panel view. The hierarchical tree in the upper section allows browsing all uploaded content. Selecting an element from it
will fill lower panels with more detailed information and its available display actions.

additional files are not processed, so the future user of the
database should process them him/herself. Such files will be
paired with the calculation CML file during its existence inside
the ioChem-BD system to provide further information on the
calculation (Figure 3).
Once the CompChem CML file is generated, a second data
flow is triggered to extract the corresponding metadata fields.
By using XSLT style sheets, we infer fields such as type of
calculation, methods used, basis set, charge, multiplicity, and
several others. From the CREATE database, we will also
retrieve additional information such as structural (which files
are involved in this upload process) and administrative (how
these files were generated). Figure 4 depicts this process that
ends building a METS compliant file of administrative,
descriptive, and structural metadata containing all aspects of
the upload.
Prior to the data storage by the CREATE module, there is a
final step aimed to extract the final geometry, a key point to
repeat the calculation if needed. This particular point sets our
computational database close to other structural databases like
the CSD (molecules)35 and for the structures of compounds in
crystallography COD.36 In the case of geometry optimizations,
a large number of geometries can appear in the same file. Again,
we rely on XSLT templates to contain the necessary logic to
retrieve the optimized geometry. The geometry is then indexed
with ChemAxon JChemBase software for future substructure
searches.37
When all these processes are completed, newly uploaded
calculations are accessible on the CREATE module via tree
browsing or search (Figure 3). At this point, users can browse
their uploaded content. Selecting a calculation opens an
auxiliary window in lower right corner with all available
actions: visualize molecule on JSmol viewer,38 view an HTML5

Figure 4. METS file generation workflow. Using XSLT stylesheets, we
extract calculation descriptive metadata fields. Together with these
fields, we append structural and administrative information to
compose a METS file that fully describes our new uploaded result.

98

dx.doi.org/10.1021/ci500593j | J. Chem. Inf. Model. 2015, 55, 95−103

Journal of Chemical Information and Modeling

Article

search filter. Results vary depending on the privileges that the
user possesses toward CREATE calculations. They are defined
by fine-grained access rules set at the user, group, and others
levels, like UNIX system file rights.
Next to JSmol visualization, another remarkable action is the
HTML resume (Figure 7). Using XSLT style sheets, ioChem-

resume of relevant calculation data (or other attaches files),
download and visualize CML, and attached files. To keep the
system’s extensibility, all actions applicable to content are
implementations of an abstract Action class that is managed via
an ActionManager object. Such a class acts also as a class loader.
This allows upgrading the system with new calculation
operations without the need to update its code, just dropping
a new Action implementation class package in the Web server
class path.
Once uploaded, it is possible to search the stored data. The
search functionality relies on standard database queries in
conjunction with the JChemBase search engine to filter its
content.37 As shown in Figures 5 and 6, users can query

Figure 5. CREATE search panel allows users to define multiple search
criteria using boolean logic. Such queries range from administrative
metadata, chemical related terms, and chemical substructures.
Figure 7. Every uploaded calculation has a group of actions associated
with it. One of them is a HTML summary that displays its most
remarkable fields. Such a summary can be customized to fulfill
researchers’ needs and to adapt to future requirements.

BD is able to generate a fully compliant HTML5 resume that
implements features such as one page presentation, all data sets
exportable to other formats, compact drop-down content,
device responsiveness, and its most valuable feature of being
fully customizable with new data fields without the need to
upgrade the platform.
Another feature delivered in HTML5 reports is the visual
representation of data. A reference to Highcharts (a Javascript
charting library)39 has been included in all generated reports.
This inclusion eases the process to convert plain data into
interactive visual elements using (among others) line, scatter, or
column charts. This inclusion behavior is easily replicable to the
innumerable third party plugins that exist today in the
chemistry field.
In addition, this report file can contain other rich content
objects such as third party plugins, navigable data tables, and
interactive graphics, among others. An example of ioChemBD’s pluggability with external tools can be observed in the
integration process done inside the HTML5 report generation

Figure 6. Search output can be narrowed by the definition of a
molecular substructure that will refine its results. A visual HTML5
molecular editor is displayed on the user’s browser to sketch fragments
or the entire molecule.

administrative and descriptive metadata fields and use a
molecular editor to sketch substructures that will be used as a
99

dx.doi.org/10.1021/ci500593j | J. Chem. Inf. Model. 2015, 55, 95−103

Journal of Chemical Information and Modeling

Article

engine with JCAMP-MOL IR Spectrum Viewer applet.40
During the development of this engine, there was a need to
include an IR viewer so that calculated vibrational frequencies
could be displayed as an additional visual field inside the
resume. To do so, a java servlet was created to convert CML
calculations to Jcamp-DX41 compliant output text by the use of
XSLT transformations. Now, calling this servlet with a
calculation ID will return its vibrational information in the
Jcamp-DX format, so appending the applet tag calling this
servlet inside our report did the job. No major code
development was required.
CREATE Publication Mechanism. Communication between
both ioChem-BD modules is currently unidirectional, from
CREATE to BROWSE modules, through a process called
“Content publication”. Publishing allows importing single
calculations or groups of them to the BROWSE module to
generate assets like reports. To complete this step, it is only
necessary to name calculations and mark them for publication.
The remaining process, REST API communications, is invisible
to the user. Because both modules are written as Java Web
services, the publication mechanism is done via servlets ,and
published files are bundled in DSpace METS SIP42 format
during its ingestion in the BROWSE module.
From this step onward, published calculations will be called
“items”. As a result of the publication process, a group of URL
handles referring to published items are presented. These links
point to public HTML pages in the BROWSE module with the
following content: (i) final calculation geometry visualization
with Jsmol, (ii) expandable summary of the item’s metadata,
(iii) summary of the most relevant data in HTML5 format, (iv)
list of downloadable content such as input files, and (v) support
files and gray literature associated with calculations. In this, final
content administrative metadata related to the purpose of the
calculation, methodology, or other relevant information from
the user acting as creator can be uploaded. Most of these
sections can be mapped to CREATE Actions as they share the
same conversion style sheets. Therefore, results share
coherency in both modules.
BROWSE Module. The BROWSE module consists of a
heavily modified version of the DSpace digital repository.18 It
has been adapted to fulfill our requirements, mainly in quantum
chemistry data representation and in external services
communication. Some workflows have been copied from the
CREATE module to have a similar behavior between them.
One of the main features in BROWSE (DSpace) instances is
that they can communicate between them using the OAI-PMH
protocol21 to share item metadata. This allows building a public
distributed network of theoretical chemistry and materials
science repositories, which will be a great advance in terms of
information socialization.
The module works by default with the Dublin Core metadata
schema,43 which is good to capture the most basic bibliographic
information about any digital asset but cannot hold the
description of quantum chemistry documents. However, this
module is versatile enough to expand its metadata schemas with
new ones, so we have created a schema focused on the
computational chemistry field. Among other interesting
features, the BROWSE module accepts browsing and searching
content,and such content can be embargoed, exported, or
syndicated depending on the users’ needs.
A notable aspect of the BROWSE module is its ability to
display supporting information and other derived chemical
reports built with the CREATE module. As a brief overview,

supporting information documents are normally composed of
one or several chemical structures (normally with the XYZ
format) from a series of related calculations. It can also contain
extra information fields such as final energies, vibrational
frequencies, spin angular momentum, etc. Supporting information documents are normally stored on heterogeneous locations
like public ftp servers, private Web servers, cloud storage
services, etc., depending on the data publication policy of each
research center. These documents are later pointed to by
journal papers as additional information related to research.
Usually, they are generated manually in a tedious, time-wasting,
and error-prone action that sometimes derivates on unportable
digital documents (one, at most two stars of five in the Open
Data Scheme).44 We try to remove such an ineffective
procedure using the supporting information generator that is
integrated inside the CREATE module and whose results are
displayed in BROWSE (Figure 8). It uses XSL-FO, an open
format object definition language, as a bridge between raw data
and multiformat output. Such a report engine is feed with CML
calculations from a user selection at the CREATE main panel
tree. After setting them in session, the user chooses to create a
new report from it. In this case, we choose Supporting
Information as report type. A fast XPath query will return

Figure 8. Supporting information report generation workflow. Starting
from a user selection of calculations, the module extracts its molecular
geometry (among other fields like final energies) to bundle them into
a single XML file. Following iterations will convert it to a XSL-FO
format and then to the user’s desired output format.
100

dx.doi.org/10.1021/ci500593j | J. Chem. Inf. Model. 2015, 55, 95−103

Journal of Chemical Information and Modeling

Article

additional fields (like final energies) that exist among these
calculations and that will (dis)activate additional report
generation options.
After setting up the report fields, the engine will extract the
final geometries and other fields from chosen calculations.
Then, they are joined into a single CML file. The next step in
report generation is to convert CML to a XSL-FO document;
with some more XSLT work, we obtain a XSL-FO document
ready to be converted based on users’ choice to any kind of
digital document such as PDF, TXT, CSV, etc.
Using a similar process, ioChem-BD is able to build a daily
growing set of reports. In this case, we can opt to generate two
types of outputs: a ready to download multiformat XSL-FO
document (similar to a supporting information report) or an
HTML5 Web page that will pop up in a new tab. This last
option is extremely versatile because it opens the door for
adding third party plugins and other dynamic content to our
report, a more powerful way to display results.
As an example of this functionality, we describe energy
reaction profile report generation. CREATE users need to
select a group of calculations and define a set of formulas that
constitute the energy steps. The report engine will build a
dynamic device-responsive HTML5 report in our browser
displaying an energy profile chart for such calculations (Figure
9).

Current developments in ioChem-BD are focused on the
publication in the BROWSE module of calculation reports. At
present, reports can only be generated in the CREATE module,
but in the near future, it will be possible to generate a public
handle inside BROWSE that points to a report generation page
that, depending on its URL parameters, outputs its results in
multiple formats.
System Adaptability and Safety Considerations.
Dynamic data definition and capture is a requirement of
today’s chemistry computational sector. Quantum software
vendors periodically release new versions of its products with
the addition of new functionalities, bug fixes, data representation changes, new chemical properties, calculation methods, or
atom basis, and on the other side, chemist software users
demand more analysis tools and higher levels of data
representation.
This constant flow of structural and representational data
changes defines a list of requirements that our software tries to
fulfill with loosely coupled data management rules. With our
customized JUMBOConverters library, we can expand our data
capture rules just by expanding the XML templates definition.
We can also modify metadata capture and data presentation to
the final user with the modification of inner XSLT style sheets.
Mastering the skills necessary to modify and expand these rules
presents a small learning curve because they are based on open
and well-documented standards. Therefore, every research
group can easily adapt its ioChem-BD instance to its
requirements without the need of an external programmer.
A user authentication mechanism has been implemented
with the Jasig CAS SSO Server.25 Its session management
service allows us to append new independent Web services in a
modular fashion without the need to implement user credential
management inside our modules.
In ioChem-BD, data processing documentation has the same
relevance as the processes it tries to describe. An outdated
documentation on a highly dynamic system as the ioChem-BD
environment will unavoidably lead to confusion. Users cannot
track down recent changes, and the reimplementation of
already existing extraction rules becomes hard to avoid. In
addition to this, such rules are defined on XML, a cryptic
language that does not help its reading unless it is converted to
a user-friendly format. These new requirements led us to
develop a toolkit that manipulates Jumbo capture templates to
build a SGML/XML DocBook fileset.47 We use it as a neutral
format bridge for its later conversion into a hierarchical group
of Web pages in WebHelp format. The documentation
generation process is triggered on every template modification
and becomes instantly accessible to all CREATE users for its
reference. This effectively avoids that the documentation
becomes outdated.
All content managed inside ioChem-BD is under access
control, even published items. In the CREATE module,
calculation content is restricted at the user/group/others levels.
In the BROWSE module, content can define fine-grained access
rules and also set content embargos depending on third party
publication requirements. Splitting the system into two
separated modules that should be installed on separated Web
servers increases the overall security of the system. The
CREATE module will hold internal research data and should be
deployed in internal Web servers with few open ports to
capture uploaded calculations from HPC and for the
publication mechanism. The BROWSE module can be moved
to a public Web server, where published items will reside and

Figure 9. Example of a dynamically generated report. On the basis of
user calculation selection and the definition of multiple energy reaction
formulas, our platform is able to build and output reaction energy
profile reports.

In terms of programming code, there is an abstract class
defined for Reports. So new classes can implement its functions,
and the ReportManager class will load them, appending new
report types dynamically with no need to alter our existing
code. Extensions to more complex outputs like R language code
snippets45 or Jmol scripts46 are envisaged.
Inside ioChem-BD, all content derived from calculations is
built under demand and then streamed to the user’s browser.
There is minimal performance loss using this dynamic
generation approach, but we enormously reduce disk space
requirements and increase data veracity, avoiding the massive
storage of formatted content that over time can become
outdated or partial.
101

dx.doi.org/10.1021/ci500593j | J. Chem. Inf. Model. 2015, 55, 95−103

Journal of Chemical Information and Modeling

Article

C02-01/BQU; CTQ2011-29054-C02-02/BQU; CTQ201127033/BQU; CTQ2012-3382/BQU; CTQ2011-23140). We
also thank MINECO for support through Severo Ochoa
Excellence Accreditation 2014-2018 (SEV-2013-0319). COST
Action CM1203 “Polyoxometalate Chemistry for Molecular
Nanoscience (PoCheMoN)”, COST Action “ECOSTBio
CM1305”, and ERC-2010-258406 are also gratefully acknowledged.

also will be referred to by their handles. The whole system
relies on HTTPS protocol for its communication among users
and modules to ensure that data is always encrypted when
transferred. There is an “additional” CAS module in charge of
user validation that uses tokens for single sign on/single sign off
session management, which greatly simplifies the session
management code and detaches it from our modules.

■

3. CONCLUSIONS
The massive use of simulation techniques in chemical research
generates huge amounts of information, known as “the Big
Data problem”. The main obstacle for managing enormous
volumes of information concerns its storage in such a way that
facilitates data mining as a strategy to optimize the processes
that allow scientists to face the challenges of sustainability,
knowledge, and the rational use of existent resources. We
created ioChem-BD (www.iochem-bd.org) as a group of
services in the cloud to manage computational chemistry
input and output files. As with other database-related projects,
the concepts underlying our platform rely on well-defined
standards, and it manages treatment, hierarchical storage, and
data recovery tools to facilitate data mining. This software
implements new methodological strategies that promote
optimal reuse of results and accumulated knowledge and that
improve researchers’ daily productivity. It automates the
extraction of relevant data and transforms numerical data into
tagged data inside its database. This platform provides tools for
the researcher in order to validate, enrich, publish, and share
information, and tools for accessing and visualizing data. Other
modules allow the automatic creation of both reaction energy
profile plots (by combining data of a set of molecular entities)
and Supporting Information files. Besides, ioChem-BD is
capable of performing kinetic analysis from reaction energy
profiles, QSSR analysis, or build data sets for screening, for
instance. Evaluation of these facilities is currently being carried
out in our groups.
The final goal is to build a new reference tool in
computational chemistry research and to fill the gap between
the generation of results and the publication of manuscripts
embedded in bibliography management and services to third
parties. Future implementations will include integration with a
semantic database by taking advantage of XSLT transformations to create data triples of every uploaded calculation.
With such information, we will be in the position to connect
our semantic data with other external data sources and to
develop a REST API to open bridges between the BROWSE
module and third party data services.44
A list of current working instances of ioChem-BD software
and a demo server are accessible at www.iochem-bd.org.

■

REFERENCES

(1) Lynch, C. Big data: How do your data grow? Nature 2008, 455,
28−29.
(2) Harvey, M. J.; Mason, N. J.; Rzepa, H. S. Digital data repositories
in chemistry and their integration with journals and electronic
notebooks. J. Chem. Inf. Model. 2014, 54, 2627−2635 DOI: 10.1021/
ci500302p.
(3) Ramakrishnan, R.; Dral, P. O.; Rupp, M.; von Lilienfeld, O. A.
Quantum chemistry structures and properties of 134 kilo molecules.
Sci. Data 2014, DOI: 10.1038/sdata.2014.22.
(4) Berners-Lee, T. The Next Web. Ted Conference. http://www.
ted.com/talks/tim_berners_lee_on_the_next_web.html (accessed
September 17, 2014).
(5) Frey, J. G.; Bird, C. L. Cheminformatics and the semantic Web:
Adding value with linked data and enhanced provenance. Wiley
Interdiscip. Rev.: Comput. Mol. Sci. 2013, 3, 465−481.
(6) Phadungsukanan, W.; Kraft, M.; Townsend, J.; Murray-Rust, P.
The semantics of chemical markup language (CML) for computational
chemistry: CompChem. J. Cheminf. 2012, 4, 15.
(7) Chen, M.; Stott, A. C.; Li, S.; Dixon, D. A. Construction of a
Robust, large-scale, collaborative database for raw data in computational chemistry: The collaborative chemistry database tool (CCDBT).
J. Mol. Graphics Modell. 2012, 34, 67−75.
(8) Adams, S.; de Castro, P.; Echenique, P.; Estrada, J.; Hanwell, M.
D.; Murray-Rust, P.; Sherwood, P.; Thomas, J.; Townsend, J. The
Quixote Project: Collaborative and open quantum chemistry data
management in the Internet age. J. Cheminf. 2011, 3, 38.
(9) AiiDA Project Home Page. http://www.aiida.net/ (accessed
November 17, 2014). Computational Materials Repository (CMR)
Home Page. http://cmr.fysik.dtu.dk (accessed September 17, 2014).
Novel Materials Discovery Repository (NoMaD) Home Page. http://
nomad-repository.eu (accessed September 17, 2014). CCSIRO
Nanostructure Data Bank Home Page. https://data.csiro.au/dap/
search?q=nanostructure (accessed September 17, 2014).
(10) The Materials Project Home Page. https://www.
materialsproject.org (accessed September 22, 2014).
(11) Hummelshøj, J. S.; Abild-Pedersen, F.; Studt, F.; Bligaard, T.;
Nørskov, J. K. CatApp: A Web application for surface chemistry and
heterogeneous catalysis. Angew. Chem., Int. Ed. 2012, 51, 272−274.
(12) World Wide Web Consortium. Extensible Markup Language
(XML) 1.0, third edition, specification. http://www.w3.org/TR/RECxml (accessed September 17, 2014).
(13) Java schema validation class, javadoc definition. http://docs.
oracle.com/javase/7/docs/api/javax/xml/validation/Validator.html
(accessed September 17, 2014).
(14) Adams, N.; Cannon, E. O.; Murray-Rust, P. Chemaxiom−An
ontological framework for chemistry in science. Nat. Proc. 2009,
DOI: 10.1038/npre.2009.3714.1.
(15) World Wide Web Consortium. XML Path Language, Version
1.0 http://www.w3.org/TR/xpath (accessed September 17, 2014).
(16) World Wide Web Consortium. XSL Transformations (XSLT),
Version 1.0, W3C Recommendation, November 16, 1999. http://
www.w3.org/TR/xslt (accessed September 17, 2014).
(17) HTML5 − A Vocabulary and Associated APIs for HTML and
XHTML. http://www.w3.org/TR/2012/CR-html5-20121217/ (accessed September 17, 2014).
(18) Smith, M.; Barton, M.; Bass, M.; Branschofsky, M.; McClellan,
G.; Stuve, D.; Walker, J. H. DSpace: An Open Source Dynamic Digital

AUTHOR INFORMATION

Corresponding Authors

*E-mail: moises.alvarez@urv.cat (M.Á -M.).
*E-mail: cbo@iciq.cat (C.B.).
Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS
Financial support for this work from the AGAUR (ref 2009
SGR 25, 2014 SGR 199, and 2014 SGR 409) of Generalitat de
Catalunya is grateful acknowledged, along with the Spanish
Ministry of Science and Innovation (project CTQ2011-29054102

dx.doi.org/10.1021/ci500593j | J. Chem. Inf. Model. 2015, 55, 95−103

Journal of Chemical Information and Modeling

Article

Repository. D-Lib Magazine 2003, http://www.dlib.org/dlib/
january03/smith/01smith.html.
(19) Apache Lucene. A high-performance, full-featured text search
engine library. http://lucene.apache.org (accessed September 17,
2014).
(20) Jmol Home Page. http://jmol.sourceforge.net/ (accessed
September 17, 2014).
(21) Lagoze, C.; Van de Sompel, H. The Open Archives Initiative:
Building a Low-Barrier Interoperability Framework. In Proceedings of
the First ACM/IEEE-CS Joint Conference on Digital Libraries, ACM,
New York, U.S.A., 2001.
(22) Gartner, R. METS: Metadata Encoding and Transmission
Standard; JISC Techwatch Report TSW; Library of Congress:
Washington, DC, October 2−5, 2002.
(23) Allinson, J.; François, S.; Lewis, S. Sword: Simple Web-service
offering repository deposit. Ariadne 2008, http://www.ariadne.ac.uk/
issue54/allinson-et-al/.
(24) HTTP over TLS Description. https://tools.ietf.org/html/
rfc2818/ (accessed September 17, 2014).
(25) Addison, M. S.; Battaglia, S.; Petro, A. Jasig CAS
Documentation. http://jasig.github.io/cas/4.0.0/index.html (accessed
September 17, 2014).
(26) Gaussian Home Page. http://jasig.github.io/cas/4.0.0/index.
html (accessed September 17, 2014).
(27) ADF Home Page. http://www.scm.com/ADF (accessed
September 17, 2014).
(28) VASP Home Page. http://www.vasp.at (accessed September 17,
2014).
(29) SIESTA Home Page. http://departments.icmab.es/leem/siesta
(accessed September 17, 2014).
(30) Turbomole Home Page. http://www.turbomole.com (accessed
September 17, 2014).
(31) Molcas Home Page. http://www.molcas.org (accessed
September 17, 2014).
(32) Orca Home Page. http://cec.mpg.de/forum (accessed September 17, 2014).
(33) JUMBOconverters. Main Project Page. https://bitbucket.org/
wwmm/jumbo-converters (accessed September 17, 2014).
(34) Murray-Rust, P.; Townsend, J.; Adams, S. E.; Phadungsukanan,
W.; Thomas, J. The Semantics of chemical markup language (CML):
Dictionaries and conventions. J. Cheminf. 2011, 3, 43.
(35) Cambridge Structural Database Home Page. http://www.ccdc.
cam.ac.uk/Solutions/CSDSystem/Pages/CSD.aspx (accessed September 17, 2014).
(36) Crystallography Open Database Home Page. http://www.
crystallography.net/ (accessed September 17, 2014).
(37) JChem Base, Chemical Interface to Relational Database Engines.
http://www.chemaxon.com/products/jchem-base (accessed September 17, 2014).
(38) JSmol, Sourceforge Project. http://sourceforge.net/projects/
jsmol/ (accessed September 17, 2014).
(39) Highcharts Home Page. http://www.highcharts.com (accessed
September 17, 2014).
(40) Hanson, R. M.; Lancashire, R. J. In JCAMP-MOL: A JCAMP-DX
extension to allow interactive model/spectrum exploration using Jmol and
JSpecView. The ACS 2013 Symposium on Exchangeable Data
Formats,American Chemical Socoety, September 11, 2013, Indiana,
IN, U.S.A.
(41) IUPAC CPEP Subcommittee on Electronic Data Standards
Home Page. http://www.jcamp-dx.org (accessed September 17,
2014).
(42) DSpace METS Document Profile for Submission Information
Packages (SIP). https://wiki.duraspace.org/display/DSPACE/
DSpaceMETSSIPProfile (accessed September 17, 2014).
(43) DCMI Metadata Terms definition Pge. http://dublincore.org/
documents/dcmi-terms/ (accessed September 22,2104).
(44) Five Star Open Data Home Page. http://5stardata.info
(accessed September 17, 2014).

(45) The R Project for Statistical Computing. http://www.r-project.
org (accessed September 17, 2014).
(46) Jmol /JSmol Interactive Scripting Documentation. http://
chemapps.stolaf.edu/jmol/docs (accessed September 17, 2014).
(47) Ortiz, I. M.; Moreno, P.; Sierra, J. L.; Manjón, B. F. Using
DocBook and XML technologies to create adaptive learning content in
technical domains. Int. J. Comput. Sci., Appl. 2006, 3, 91−108.

103

dx.doi.org/10.1021/ci500593j | J. Chem. Inf. Model. 2015, 55, 95−103