This is an open access article published under a Creative Commons Non-Commercial No Derivative Works (CC-BY-NC-ND) Attribution License, which permits copying and redistribution of the article, and creation of adaptations, all for non-commercial purposes.
Article Cite This: ACS Omega 2019, 4, 86−94
http://pubs.acs.org/journal/acsodf
Substance-Based Bibliometrics: Identifying Research Gaps by Counting and Analyzing Substances Robert Tomaszewski* California State University, Fullerton, 800 North State College Blvd, Fullerton, California 92831 United States
Downloaded via 178.159.100.176 on January 12, 2019 at 14:41:38 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
S Supporting Information *
ABSTRACT: Identifying research gaps and generating research questions are often a first step in developing ideas for writing a research paper or grant proposal. The concept of substance-based bibliometrics uses the counts of substances in the scientific literature to better understand, assess, and clarify the state and impact of information in the chemical sciences. Connecting substances indexed to specific bioactivity or target indicators can lead to assessing the biochemical, biological, and medicinal relevance of substances as well as developing ideas for expanding drug design and discovery through identifying and modifying the structural features of molecules. This study uses Chemical Abstracts through the SciFinder database to count for the occurrence of substances in the scientific literature. The study sets out search strategies for discovering potential research gaps and new ideas through visualization of chemical structures with known bioactivity and target indicators. The author recommends that subject librarians integrate research gap training in their bibliographic instruction classes, particularly to upper-level undergraduate and graduate chemistry students. literature review is often the first step to finding all literature and information about a research topic. Browsing through review papers is one approach for an overview, perspective, and understanding of a research topic as well as finding leading researchers in a specific field.1 It is also important to continuously analyze recent review papers to keep abreast on the current state of a research area. Systematic reviews are types of literature reviews containing many research studies that dwell deep into the literature with an objective to analyze the trends and changes in a field of study.2,3 When conducting a systematic review, there are often protocols involved in deciding what information is included and not included as well as more evaluative statements made about the studies. In more traditional review articles, the tone is more neutral and often there is no indication of how the articles referenced were chosen. Searching peer-reviewed articles for information on future research (i.e., occasionally found in the conclusion section or under a separate heading like “Future Work”) offers another approach to identifying gaps in the literature. The research methodology found in articles may also spark new ideas for alternative approaches to a problem. The funding or acknowledgment section can provide information about the grant source for the research and could be an indicator where to submit similar research ideas. Searching through dissertations and
1. INTRODUCTION A research gap is a question or a problem in a subject field that has been answered incompletely or insufficiently or that has not been answered at all. Research gap spotting and generating research questions are often a first key step to identifying and developing ideas for writing a research grant proposal or paper. The initial research questions scientists often ask include how, who, what, which, and why: How far back does the literature extend on a scientific topic, element, substance, or reaction? Who are the leading scientists working in a certain research field? What are the potential applications for specific chemical structures? Which substances show bioactivity? Which structural features are crucial for specific bioactivity or target affinity? Which reactions do not work? Why do certain functional groups or elemental species affect bioactivity? Many science degree programs require students to write and defend a research proposal; however, the process for discovering gaps in the literature seldom receives any formal training. As a result, this creates an outreach opportunity for science librarians to become more involved with researchers and students through communication modes such as bibliographic instruction and online guides. 1.1. Research Gaps: Literature Orientated. Searching the literature through different document types is always a good starting point for finding information about a research topic and its current state. To this regard, although many journals are dedicated to publishing review papers solely, high profile journals generally answer and solve a problem. Performing a © 2019 American Chemical Society
Received: August 28, 2018 Accepted: November 5, 2018 Published: January 2, 2019 86
DOI: 10.1021/acsomega.8b02201 ACS Omega 2019, 4, 86−94
ACS Omega
Article
browsing laboratory notebooks can provide more information on the progress of research. Various reports such as content analysis reports, citation analysis reports, and meta-analysis reports occasionally contain information on ideas for further research studies. 1.2. Research Gaps: Researcher Orientated. Communicating with experts in a field is another approach in understanding the state of a research area. Attending subjectspecific conferences conducted by experts in a research field provides an opportunity to network and ask questions about extending research and debating ideas for future studies. In particular, conferences offer opportunities for dialogue at seminars, workshops, panel discussions, oral presentations, poster sessions, and luncheons as well as various informal gatherings. Visiting websites of leading researchers can provide ongoing information about their research and the challenges ahead. Connecting and corresponding with researchers through email, social media, or face-to-face are one direct approach for input about a research idea. Communication with the dissertation author may provide more direct insight into the state of the research in the laboratory. On the other hand, some scientists may be reluctant to discuss their ideas from being scooped with their research. Moreover, because of the publish or perish nature in academia, some researchers feel it is best to publish selected results in small amounts or at times when they feel they have fully researched the subject area. Sometimes, it is a matter of just communicating at the right time and right place. 1.3. Research Gaps: Field Orientated. Identifying and analyzing the state of a research topic or field through online tools can provide guidance and direction on the progress of research and development. General online resources, such as Web of Science, Scopus, Google Scholar, Essential Science Indicators, and Google Trends, and subject-specific resources, such as SciFinder, Reaxys, PubChem, PubMed, among many others, provide access to finding information about related or most cited documents, and in many cases, the popularity in a research field. Bibliometrics through collecting, counting, and analyzing publication and citation information can further be used to address trends in the literature. The use of different search strategies with databases can provide past and present information on the state of research about a scientific topic as well as background data on substances.4−6 Schummer (1997) used Chemical Abstracts, Beilstein, Gmelin, and other chemical handbooks to show an essentially exponential quantitative growth of substances between the years 1800 and 2000 with significant deviations stemming from world wars and a “catching up phenomena” during postwar periods.7 Schummer (1997) also pointed out that “there is still no saturation discernible after 200 years of exponential growth of substances.”8 Barth and Marx (2012, 2013) have introduced the idea of compound-based bibliometrics by searching for compounds containing rare-earth elements in Chemical Abstracts and linking to their corresponding publications.9,10 They stated that “the method can be applied to analyze large amounts of compounds in combination with the corresponding chemical concepts, to identify gaps in research, and hence open the door to new research in well-described compound-based areas.”10 It was concluded that “the compound-based bibliometric concept can easily be extended to organic chemistry by searching molecular substructures rather than element combinations, or to biochemistry by searching protein or nucleic sequences.”10
With the emergence of the internet and chemical databases such as SciFinder, Reaxys, ChEMBL, and PubChem, it has now become possible to quickly visualize and compare the structural makeup of property-specific substances and brainstorm ideas for continued research work.11 To this regard, SciFinder is arguably the most comprehensive database covering chemical literature (Table 1).12−18 SciFinder further allows for substance searching Table 1. Coverage and Content Information in SciFinder SciFinder coverage and update
indexing
searchable information
CAplus (references, 1800s to present, updated daily)a CAS REGISTRY (chemical substances, 1800s to present, updated daily) CASREACT (reactions, 1840 to present, updated daily) CHEMCATS (chemical suppliers, 2013 to present, updated weekly) CHEMLIST (regulated chemicals, 1980 to present, updated weekly) MARPAT (Markush structures, 1961 to present, updated daily) CIN (chemical industry notes, 1974 to present, updated weekly) MEDLINE (1946 to present, updated daily) over 50 000 scientific journals, publication dates go back to 1907, but pre-1907 content (>224 000 records) is also made accessible back to the 1800s over 47 million records chemistry and other science-related research records, such as journals, patents (63 patent authorities), reports, books, conference proceedings, dissertations, and synthetic preparations over 144 million organic and inorganic substances over 67 million DNA and protein sequences over 8 billion property values, data tags, and spectra millions of single and multi-step reactions over 348 000 inventoried/regulated substances over 1.1 million Markush structures over 1.7 million record industry notes bioactivity and target indicators CAS registry number chemical structures experimental and predicted property data Markush molecular formula reactions text or numeric
a
CAplus Core Journal Coverage List consists of around 1500 journals from which abstracts are added and bibliographic, substance, and reaction information are indexed within a few days of publication, (https://www.cas.org/support/documentation/references/ corejournals).
within the chemical literature.19 SciFinder has assigned physical property data as well as bioactivity and target indicators to substances through the reported literature (Table 2).20−22 Bioactivity indicators are a set of controlled vocabulary used to report a particular biological activity in the literature (e.g., antiinflammatory agents, antitumor agents, cardiovascular agents, enzyme inhibitors, immune agents, nervous system agents, reproductive control agents, wound healing promotors).16,20 Target indicators are a set of controlled vocabulary used to report particular biological targets in the literature (e.g., albuminoids, apoptosis-regulating proteins, blood-coagulation factors, cytokines, enzymes, globulins, glycoproteins, protease inhibitors, RNA formation factors).16,20 These indicators are reminiscent of National Therapeutic Indicators. A complete list 87
DOI: 10.1021/acsomega.8b02201 ACS Omega 2019, 4, 86−94
ACS Omega
Article
Table 2. Substance Indicator and Property Search Options in SciFinder indicator or property type bioactivity indicator target indicator experimental property predicted property
searchable options over 260 bioactivity indicators (e.g., anti-inflammatory agents, antitumor agents, cardiovascular agents) over 5800 target indicators (e.g., albuminoids, apoptosisregulating proteins, blood-coagulation factors) 13 different experimental property data (e.g., boiling point, magnetic moment, melting point, tensile strength) 21 different predicted property data (e.g., bioconcentration factor, density, molar solubility, pKa, vapor pressure)
of these indicators is however not publicly available.23 The presence or absence of bioactivity and target indicators can be used to identify potential areas for further research and drug design. By comparing the number of substances to an associated bioactivity or target indicator, scientists can analyze and evaluate relevant structural features to extend and develop research ideas. Potential research gaps and new ideas can be singled out through visual examinations and assessments of chemical structures with known bioactivity and target indicators. This study uses titanocene dichloride, (η5-C5H5)2TiCl2 (CAS registry number: 1271-19-8), as a model molecule to identify and compare substances that show antitumor activity using different search strategies with the SciFinder database. The substance searches conducted in the study include the following: • Exact structure, substructure, and similarity (a score of 90 as a cutoff point) search for titanocene dichloride (Table 3 and Figure 1). The search is used to compare and analyze biological activity information obtained from the three different search types.
Figure 1. Substance searches for titanocene dichloride with corresponding bioactivity indicator identified as antitumor agents.
Table 4. Similarity Searches (Score ≥90) for Titanocene Dichloride and Similar Structures Using SciFinder
Table 3. Bioactivity and Target Indicators for Titanocene Dichloride from an Exact Structure, Substructure, and Similarity (Score ≥90) Search Using SciFinder search type exact Structure substructure
bioactivity indicators (number of substances)
target indicators (number of substances)
antitumor agents (9) anti-infective agents (1) antitumor agents (315)
enzymes (1)
anti-infective agents (51) anti-inflammatory agents (27) immune agents (27) antiproliferative agents (5) enzyme inhibitors (4) reproductive control agents (2) similarity (score ≥90)
antitumor agents (37)
apoptosis-regulating proteins (13) ubiquitin (13) enzymes (12) transport proteins (7) globulins (5) glycoproteins (5) hemoproteins (2) RNA formation factors (2) enzymes (1)
2. RESULTS AND DISCUSSION 2.1. Substance Counts by Search Type. According to Miller (2002), “an exact-match search can be thought of as looking up a complete word in a dictionary. A substructure search is analogous to a wild-carded text search, and a similarity search resembles a “sound-like” search.”24 An exact structure search retrieves the substance exactly as drawn and component systems. A substructure finds the exact substance and the searched structure embedded within other substances. A similarity search finds similar structures based on similarity scores from a highest (most similar) to lowest (least similar) value. In SciFinder, a similarity search uses a Tanimoto algorithm to retrieve all structures based on similarity.25,26
anti-infective agents (3)
• Similarity searches on titanocene dichloride through addition and replacement of ligands around the central titanium metal (Table 4). The search uses the idea of ligand substitution to analyze and evaluate substances for biological activity. • Substructure searches on titanocene dichloride through substitution of functional groups on the Cp ligand (Table 5). The search uses the idea of ligand modification to analyze and evaluate substances for biological activity. 88
DOI: 10.1021/acsomega.8b02201 ACS Omega 2019, 4, 86−94
ACS Omega
Article
Performing an exact structure search in SciFinder does not allow for variations in atoms (e.g., generic groups or R-groups); however, variable bonds are allowed (e.g., unspecified bonds). On the other hand, similarity searches cannot be done with structures that contain R-groups, variables, repeating groups, variable attachment, multiple fragments, and stereo bonds.27,28 A substructure search is generally performed when searching for specific structural or atom requirements in a chemical structure, whereas a similarity search is used for exploring similar substances with greater variation in the literature.27,28 The results from an exact structure search are redundant with both the substructure and a similarity search because they form a subset of each of the latter two searches. However, by separating the searches, the results can be refined accordingly for focusing on the substance structure (e.g., substances with the discrete structures or embedded structures) as well as limit the number of substances in the set for easier visualization, interpretation, and comparative analysis. A similarity search also ensures finding substances within a component system such as copolymers, mixtures, and salts even though it may not be present as an exact form or contain a CAS registry number. Performing an exact structure or similarity search separately yields fewer substances that are structurally similar. Fewer substance counts can facilitate easier and better visualization for comparing closely related structures (e.g., viewing 9 or 37 substances is easier than 315 substances, Figure 1). Further, an exact structure search retrieves multicomponent substances, where the searched molecule is present and discrete. This can lead to identifying research gaps based on modifying substance components rather than the searched substance. The type of bioactivity and target indicators found from an exact structure, substructure, and similarity (score ≥90) search using (η5-C5H5)2TiCl2 are shown in Table 3. It is evident from Table 3 that antitumor agents are the most common type of bioactivity indicator from each search type. In addition, the majority of different target indicators were identified from a substructure search. Figure 1 shows the counts of substances identified as antitumor agents for (η5-C5H5)2TiCl2 to each search type. An exact structure search for (η5-C5H5)2TiCl2 yielded 190 substances where the majority were multicomponent systems, whereas others were labeled with 13C or D on the rings. From the 190 set of substances, two classes of bioactivity indicators were identified with nine substances as antitumor agents (Table 3). One substance was the parent compound, and eight substances were multicomponents in the form of copolymers. A slight modification of the backbone of the organic fragment of the polymer could inspire new research. Further, accessing articles to these substances can provide synthesis information as well as the funding sources to submit research proposals. A substructure search for (η5-C5H5)2TiCl2 yielded 3224 substances from which were identified seven different bioactivity indicators (Table 3). A total of 315 substances were classified as antitumor agents (Figure 1). Retrieving references to each indicator can be found by going directly to the substance “handbook format” table of information by clicking on the “CAS number” or via view “Substance Detail” (Figure 2). Information such as specific enzymes can be located from this table (e.g., target indicators for titanocene dichloride are enzymes to which is stated glutaminase with access to 13 references). Using the variable function from the structure editor, a substructure search conducted on (η5-C5H5)2TiX2 (where X = any halide), yielded 5728 substances from which identified a slight increase of 349
Table 5. Substructure Searches for Analogues of Titanocene Dichloride Using SciFinder
The algorithm uses statistical analysis to compare the structure drawn to all other structures in the database. The similarity scores have a scale range of 0−100 and are based on CAS structure descriptors such as element composition, atom count, ring count, atom sequence, bond sequence, and degree of connectivity.26 A similarity search could result in retrieving substances with functional groups or bonding arrangements that may not be obvious to the researcher. The most similar structures are identified as a higher score (e.g., similarity score of ≥99 contains the most similar structures). According to SciFinder (2005), substances with a similarity score above 60 will usually be displayed.26 89
DOI: 10.1021/acsomega.8b02201 ACS Omega 2019, 4, 86−94
ACS Omega
Article
Figure 2. Accessing bioactivity and target indicators from a substance table in SciFinder.
adding or removing terms in a Boolean search allows to narrow or broaden the resulting search to retrieve the most relevant hit set. This building search process provides a strategy to identify a parent species with a bioactivity or target indicator that holds the most promise for further investigation. To this regard, the obvious choice for (η5-C5H5)2TiCl2 would be to remove or add more Cp and chloro groups and deduce the best combination of ligands around the titanium center for the highest counts of substances with antitumor activity. From Table 4, it can be deduced that two Cp rings and one or two chloro groups on titanium result in the highest number of antitumor substances. This hints that modification of these structural features could lead to other substances with potential biological activity and sets out an opportunity for more synthetic ideas. Likewise, the half-sandwiched species, CpTiCl3, was found to have one substance identified as an antitumor activity. Consequently, an opportunity for further research could be envisioned through replacement or modification of the Cp rings and ancillary ligands. This in turn could lead to improved antitumor activity. A substructure or similarity search can be used to identify different metals, in addition to isotope elements and multicomponent systems such as counterions. Furthermore, trends from the periodic table (e.g., elements with the most similar properties are located in the same group) can be used to identify other elements for replacement (i.e., Ti is in the same group as Zr and Hf). This can bring forth ideas for extending the synthesis to other potential substance analogues exhibiting bioactivity.
substances as antitumor agents (i.e., one of seven different bioactivity indicators). A similarity search (score ≥90) for (η5-C5H5)2TiCl2, yielded 711 substances from which were identified 37 substances as antitumor agents and 3 substances as anti-infective agents (Table 3). By identifying the bioactivity and target type for each structure, drug discovery scientists can obtain supportive evidence and guidance for further studies. According to Garritano (2013), “it is also possible to identify substances that do not have indexed bioactivity or target indicators though they may be structurally similar to known compounds. These substances might be chosen for further research or investigation as to whether they would have similar, as yet undiscovered, biological activity or not.”21 Others have further stated that “these bioactivity and target indicators guide drug discovery scientists to new uses for known drugs, possible side effects, and the original literature where pharmaceutical information was reported.”22 2.2. Substance Counts by Ligand Addition and Replacement. A structural comparison for (η5-C5H5)2TiCl2 to other compounds through a similarity search (score ≥90) is shown in Table 4. The strategy in Table 4 was to deduce the assembly of the Cp rings and chloro groups needed for bioactivity. This can be performed by simply adding or replacing functional groups or stereochemistry around the molecule, followed by analyzing for the number of substances indexed to a specific type of bioactivity or target indicator. This type of refining search strategy is much like a keyword search, where 90
DOI: 10.1021/acsomega.8b02201 ACS Omega 2019, 4, 86−94
ACS Omega
Article
Figure 3. Refining a substance set by property data in SciFinder.
2.3. Substance Counts by Ligand and Chemical Structure Modification. Table 5 shows the results of substructure searches for analogues of titanocene dichloride indexed as the antitumor activity. Strategies and rationales for ligand substitution and modification can be considered when determining how to approach a substance search. Optimizing drug metabolism and pharmacokinetics can occur through solubility, stability, side effects, steric hindrance (bulkiness) and dimension, electronic effects (resonance and intrinsic inductive effects), and stereochemistry.29−31 Functional groups can have an electronic, solubility, and a steric effect on the substance. To this extent, the cytotoxic activity of substances may well be correlated with their structure.32 The unique properties of transition metals can also affect the activity.33−35 As a result, various rationales can be drawn when deciding how to modify a structure, such as the influence of a specific group to help with water and lipid solubility and prevent unwanted side effects; adding polar groups to increase polarity yet decrease hydrophobicity; bulkiness and electronic effects from functional groups to increase the chemical and metabolic stability of a drug [e.g., changing the position of the functional group on the phenyl ring (i.e., para, ortho, and meta positions) such as an ortho group could act as a steric shield and hinder hydrolysis]; maintaining an optimized ratio of ionized to unionized drug species by controlling the pKa through functional groups; and masking groups known for toxicity or side reactions. The strategy of incorporating a group onto a Cp ring to increase bulkiness and electronic effects could be used to enhance bioactivity or linking two Cp rings to improve hydrolytic stability. Some researchers have stated that incorporating polar electron-withdrawing groups on the Cp ring could help with the solubility as well as improve antitumor activity from increasing the Lewis acidity of the Ti(IV) center to enhance binding to DNA.36,37 Sometimes, it is a matter of trial and error to see which groups or structural modifications result in the highest counts of substances with bioactivity. Moreover, from the substance counts showing activity, a sense of structural direction
can emerge. This type of refining search strategy using functional group modification can be envisioned like a truncation or wild card search where slight modification of the search term brings up the highest number and most relevant hits. One strategy for ligand modification in Table 5 was to add a group to the Cp ring that would affect solubility and then alter the group slightly to see if the effect would increase or decrease the number of substances with antitumor activity. This effect was also examined by incorporating unreactive methoxy groups. The counts of antitumor substances were further analyzed by introducing chirality and linking two Cp rings. Replacing phenyl with a benzyl group attached to one Cp increases the number of active substances from zero to 74. Similarly, placing two benzyl groups on the same Cp ring compared to one benzyl group on each Cp ring increases the count of antitumor substances from zero to 62. A strategy to determine the best structurally fit active substance in Table 5 is to calculate the gap count between the total number of substances and the number of substances with activity (i.e., so long as there are substance counts with activity). Generally, the smaller the gap count, the more promise the active substance holds. Hence, over 45 percent of all substances with the methoxy groups has antitumor activity, and 60 percent of all substances from the ansa-bridged ligand shows antitumor activity. Thus, through modification of the structural features on the Cp ligand (e.g., ansa-bridged Cp ligand, electron-donating or electron-withdrawing groups, chirality), scientists can deduce and design the structural makeup of substances that may hold promise for potential or improved bioactivity. 2.4. Substance Set Refining by Property Values. SciFinder allows to search and to refine by experimental and predicted property information (Table 2). Substance sets can be refined by physical property value from the refine menu and selecting a property value such as H donors, H acceptors, molecular weight, and solubility, among other values within a specified range. As shown in Figure 3, selecting “Property Value” in SciFinder allows to plug-in a range for two experimental property values as well as 21 different predicted property values. 91
DOI: 10.1021/acsomega.8b02201 ACS Omega 2019, 4, 86−94
ACS Omega
Article
Selecting “Property Availability,” followed by “Any Selected Experimental Property” allows to retrieve any combination of the 13 experimental property values without specifying a range. The range for all substance properties (experimental and predicted) can be specified from the “Property” search option in SciFinder; however, this is not limited to the substance set, but all substances within the database. Refining the substance sets by specific property data can be an important factor for developing drugs such as the drug-likeness of a substance outlined by Lipinski’s rule of five (i.e., no more than 5 hydrogen donors, no more than 10 hydrogen acceptors, molecular weight less than 500, and octanol−water partition coefficient less than 5)38 or the rule of three for fragment-based drug discovery (i.e., no more than 3 hydrogen donors, no more than 3 hydrogen acceptors, molecular mass less than 300, and octanol−water partition coefficient no more than 3).39
5. CONCLUSIONS The concept of substance-based bibliometrics uses the counting of elements or compounds rather than publications and citations in the scientific literature to better understand, assess, and clarify the state and impact of information in the chemical sciences. The assignment of specific bioactivity and target indicators to millions of substances in commercial and public databases provides an opportunity to identify gaps in drug design and discovery. With the knowledge that such information is right at our fingertips, it is hoped that scientists will become inspired to utilize and apply these searching tools and strategies to their research. 6. METHODOLOGY The SciFinder database is a product produced by Chemical Abstracts Service (CAS), a division of the American Chemical Society (ACS). The coverage and content in SciFinder is shown in Table 1.12−18 The Chemical Abstracts Plus (CAplus) file of SciFinder contains millions of chemistry and other sciencerelated research records. The registry numbers of substances are included in the records in the Chemical Abstracts file of articles that employ or describe those substances. Each substance in the CAS Registry file is linked to available literature references, spectra, experimental property data, commercial availability synthesis and reactions, CA index names, and regularity information to the substance. CAS has further assigned bioactivity and target indicators to substances for assessing their biological characteristics (Table 2).20−22 Bioactivity and target indicators are identified to substances in the CAS REGISTRY database through documents in the CAplus database. These indicators are applied to a substance’s record when at least one paper is indexed as attributing a specific activity to that substance. All SciFinder searches in this study were performed between July and September 2018. Titanocene dichloride was used to compare the substance counts from an exact structure, substructure, and similarity (score of 90 as cutoff point) search. To this regard, from the “Explore” menu in SciFinder, the “Substance Identifier” option query was selected. The name of the substance, titanocene dichloride, was typed in and searched. Titanocene dichloride was moved to the chemical editor by hovering over the chemical structure and selecting “Click for more options” (double arrow), followed by selecting “Explore by Structure” and “Chemical Structure.” This copied the substance into the chemical editor which in turn allowed it to be further modified to other substances, as shown in Tables 4 and 5. It is also possible to draw the substances from scratch in the chemical editor or draw it using chemical drawing software such as ChemDraw and import it as a molfile. Starting from a name or molecular formula search, followed by copying to the chemical editor can sometimes be easier, particularly with drawing more complex groups such as η3 or η5 ligands as well as leading to fewer drawing errors. By respectively selecting exact structure, substructure, and similarity, the database retrieved all relevant substances. The “Analyze by” dropdown option from the “Analyze” menu allowed to select different options such as “Bioactivity Indicators” or “Target Indicators.” From the bioactivity indicator, all substances classified as “Antitumor agents” were selected. By clicking on the “Show More” option, followed by “Apply,” the user can select multiple indicators and retrieve the respective substances of interest for viewing.
3. APPLICATIONS IN TEACHING It is recommended that subject librarians integrate research gap training in their bibliographic instruction classes. This can be accomplished when demonstrating a search for a substance using SciFinder in the classroom, followed by analyzing the substance set by specific bioactivity or target indicator. At this point in the class, the concept of the research gap could be introduced by asking students for potential ideas and reasons for modifying the structural features in the resulting substance set to extend the research. That way, the research gap message is short and sweet, allowing students to critically think. From personal experiences, this type of approach has students engaged in creative brainstorming and sharing of ideas. Synthesizing a new substance could involve something as simple as introducing or removing a methyl group into the known substance. Moreover, given that the experimental procedure is already available for known substances, it could just be a matter of adding or changing a simple group in the reactant that would result in a new substance. Introducing the research gap is especially useful to chemistry students who are required to write a research proposal as part of their degree program or come up with a quick project idea for a synthetic organic or inorganic laboratory class. 4. FUTURE WORK Subscription-based resources such as SciFinder are seamless, user-friendly, and maintain high-quality control with the information content. These features make it all the more suited to use with the novice student researcher. However, not all information is available solely from one resource. In addition, some resources contain data that are geared toward a specific science discipline. To this regard, other subscription-based resources such as the Reaxys Medicinal Chemistry40 database also contain millions of bioactivity data points and thus worthy for future study. In addition, publicly accessible resources such as PubChem41,42 and ChEMBL43 could be compared as they provide searching capabilities to large volumes of substances and easy access to curated and annotated data. Comparing substance-based bibliometrics between commercially and publicly available resources would, therefore, be useful for further study.44,45 Future work could also explore the bioactivity and target indicators from the elements of the periodic table (elemental-based bibliometrics)46 and reaction chemistry (reaction-based bibliometrics) for applications to drug synthesis. 92
DOI: 10.1021/acsomega.8b02201 ACS Omega 2019, 4, 86−94
ACS Omega
Article
(8) Schummer, J. Scientometric studies on chemistry II: Aims and methods of producing new chemical substances. Scientometrics 1997, 39, 125−140. (9) Barth, A. Chemical bibliometrics; Chemistry World, April 25, 2013. https://www.chemistryworld.com/opinion/chemical-bibliometrics/ 6099.article (accessed Aug, 2018). (10) Barth, A.; Marx, W. Stimulation of ideas through compoundbased bibliometrics: counting and mapping chemical compounds for analyzing research topics in chemistry, physics, and materials science. ChemistryOpen 2012, 1, 276−283. (11) Wang, H.; Yin, Y.; Wang, P.; Xiong, C.; Huang, L.; Li, S.; Li, X.; Fu, L. Current situation and future usage of anticancer drug databases. Apoptosis 2016, 21, 778−794. (12) Chemical Abstracts Service. SciFinder, What is SciFinder? 2018. https://www.cas.org/products/scifinder (accessed Aug, 2018). (13) Chemical Abstracts Service. SciFinder, Database and content training documentation, 2018. https://www.cas.org/support (accessed Aug, 2018). (14) Chemical Abstracts Service. SciFinder, Training: Need to knowStructure searching, 2018. http://support.cas.org/training/scifinder/ need-to-know-structure-searching (accessed Aug, 2018). (15) Chemical Abstracts Service. SciFinder, How to... Create a substance answer, 2014. http://download.cappchem.com/Tutorial/ SciFinder-Substance.pdf (accessed Aug, 2018). (16) Chemical Abstracts Service, American Chemical Society. ReferencesCAplus-Worldwide coverage of many scientific disciplines all in one source, 2018. https://www.cas.org/support/ documentation/references (accessed Aug, 2018). (17) Chemical Abstracts Service, American Chemical Society. CAplus-Pre-1907 coverage, 2018. https://www.cas.org/support/ documentation/references/capluspre1907 (accessed Aug, 2018). (18) Chemical Abstracts Service, American Chemical Society. CAS REGISTRYThe gold standard for chemical substance information, 2018. https://www.cas.org/support/documentation/chemicalsubstances (accessed Aug, 2018). (19) Chemical Abstracts Service. SciFinder training videos: Structure searching, 2018. https://www.cas.org/support/training/scifinder/ structure-search (accessed Aug, 2018). (20) ZBChem News. Zusatzinfos zum SciFinderUpdate vom Dezember 2011, 2012. https://zbchemnews.wordpress.com/2012/ 01/05/zusatzinfos-zum-scifinder-update-vom-dezember-2011/ (accessed Aug, 2018). (21) Garritano, J. R. Evolution of SciFinder, 2011-2013: New features, new content. Sci. Technol. Libr. 2013, 32, 346−371. (22) Schenck, R. J.; Zapiecki, K. R. Back to the Future: CAS and the Shape of Chemical Information To Come. In The Future of the History of Chemical Information; McEwen, L. R., Buntrock, R. E., Eds.; ACS Symposium Series; ACS Publications: Washington, DC, 2014; Vol. 1164, Chapter 9, pp 156−157. (23) Chemical Abstracts Service. Personal communication, 2018. (24) Miller, M. A. Chemical database techniques in drug discovery. Nat. Rev. Drug Discovery 2002, 1, 220−227. (25) Willett, P.; Barnard, J. M.; Downs, G. M. Chemical similarity searching. J. Chem. Inf. Comput. Sci. 1998, 38, 983−996. (26) SciFinder. Similarity search overview, 2005. http://202.127.145. 151/siocl/scifinder/SF_Help/sf_only/tanimoto2.htm (accessed Aug, 2018). (27) Currano, J. N. Searching by Structure and Substructure. In Chemical Information for Chemists: A Primer; Currano, J. N., Roth, D. L., Eds.; The Royal Society of Chemistry: Cambridge, U.K., 2014; Chapter 5, pp 109−145. (28) Ridley, D. D. Information Retrieval: SciFinder, 2nd ed.; Wiley: Hoboken, NJ, 2009. (29) Köpf-Maier, P.; Köpf, H. Transition and main-group metal cyclopentadienyl complexes: preclinical studies on a series of antitumor agents of different structural type. Struct. Bonding (Berlin) 1988, 70, 103−185.
A similarity search (score of 90 as cutoff point) for comparison analysis was performed for “stable titanocene,” (C20H20Ti2), searched via “Molecular Formula”, whereas titanocene monochloride, titanocene trichloride, and titanium tetrachloride were searched via “Substance Identifier.” A substructure search for comparison analysis was performed by modifying titanocene dichloride in the structure editor and selecting the substructure option. The Supporting Information provides screenshots of the steps involved with searching SciFinder.
■
ASSOCIATED CONTENT
S Supporting Information *
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acsomega.8b02201.
■
Screenshots of methodology (explore by substance identifier; export structure to the chemical editor; searching the chemical structure by exact structure, substructure, or similarity; analyzing chemical structures; analyzing structures by bioactivity and target indicators; selecting bioactivity indicators; and selecting similarity score (PDF)
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected]. Phone: 657-278-2976. ORCID
Robert Tomaszewski: 0000-0001-6916-1265 Notes
The author declares no competing financial interest.
■
ACKNOWLEDGMENTS I thank the reviewers for their very insightful and intelligent suggestions with the manuscript. I also thank Marissa Medeiros for help with designing the graphical abstract.
■
REFERENCES
(1) Müller-Bloch, C.; Kranz, J. A framework for rigorously identifying research gaps in qualitative literature reviews. Proceedings of the 36th International Conference on Information Systems, Fort Worth, Texas, 2015. https://pdfs.semanticscholar.org/89e6/ a54cbe7240488d88ce49b51fc83c7186d564.pdf (accessed Aug, 2018). (2) Robinson, K. A.; Saldanha, I. J.; Mckoy, N. A. Development of a framework to identify research gaps from systematic reviews. J. Clin. Epidemiol. 2011, 64, 1325−1330. (3) Robinson, K. A.; Saldanha, I. J.; Mckoy, N. A. Frameworks for determining research gaps during systematic reviews. Methods Future Research Needs Report No. 2. AHRQ Publication No. 11-EHC043-EF; Agency for Healthcare Research and Quality: Rockville, MD, 2011. https://www.ncbi.nlm.nih.gov/books/NBK62478/ (accessed Aug, 2018). (4) Tomaszewski, R. The concept of the imploded boolean search: A case study with undergraduate chemistry students. J. Chem. Educ. 2016, 93, 527−533. (5) Wagner, A. B. Searching coordination and organometallic compounds in SciFinder. Issues Sci. Technol. Librarian. 2011, 64. DOI: 10.5062/F4G44N6W http://www.istl.org/11-fall/tips.html. (6) Wagner, A. B. Searching inorganic substances in SciFinder. Issues Sci. Technol. Librarian. 2011, 64. DOI: 10.5062/F4QJ7F79 http:// www.istl.org/11-winter/tips.html. (7) Schummer, J. Scientometric studies on chemistry I: The exponential growth of chemical substances, 1800-1995. Scientometrics 1997, 39, 107−123. 93
DOI: 10.1021/acsomega.8b02201 ACS Omega 2019, 4, 86−94
ACS Omega
Article
(30) Metabolism, Pharmacokinetics and Toxicity of Functional groups: Impact of Chemical Building Blocks on ADMET; Smith, D. A., Ed.; Royal Society of Chemistry: Cambridge, U.K., 2010. (31) Dörwald, F. Z. Lead Optimization for Medicinal Chemists: Pharmacokinetic Properties of Functional Groups and Organic Compounds; Wiley-VCH: Weinheim, Germany, 2012. (32) Kerns, E. H.; Di, L. Drug-like Properties: Concepts, Structure Design and Methods from ADME to Toxicity Optimization, 2nd ed.; Elsevier: Amsterdam, 2016. (33) Gómez-Ruiz, S.; Maksimović-Ivanić, D.; Mijatović, S.; Kaluđerović, G. N. On the discovery, biological effects, and use of cisplatin and metallocenes in anticancer chemotherapy. Bioinorg. Chem. Appl. 2012, 2012, 140284. (34) Loza-Rosas, S. A.; Saxena, M.; Delgado, Y.; Gaur, K.; Pandrala, M.; Tinoco, A. D. A ubiquitous metal, difficult to track: towards an understanding of the regulation of titanium (IV) in humans. Metallomics 2017, 9, 346−356. (35) Ndagi, U.; Mhlongo, N.; Soliman, M. Metal complexes in cancer therapyAn update from drug design perspective. Drug Des. Devel. Ther. 2017, 11, 599−616. (36) Boyles, J. R.; Baird, M. C.; Campling, B. G.; Jain, N. Enhanced anti-cancer activities of some derivatives of titanocene dichloride. J. Inorg. Biochem. 2001, 84, 159−162. (37) Gao, L. M.; Hernández, R.; Matta, J.; Meléndez, E. Synthesis, Ti (IV) intake by apotransferrin and cytotoxic properties of functionalized titanocene dichlorides. J. Biol. Inorg. Chem. 2007, 12, 959−967. (38) Lipinski, C. A.; Lombardo, F.; Dominy, B. W.; Feeney, P. J. Experimental and computational approaches to estimate solubility and permeability in drug discovery and development settings. Adv. Drug Deliv. Rev. 1997, 23, 3−25. (39) Congreve, M.; Carr, R.; Murray, C.; Jhoti, H. A “rule of three” for fragment-based lead discovery? Drug Discovery Today 2003, 8, 876− 877. (40) Elsevier. Reaxys Medicinal Chemistry, Empowering medicinal chemists with accurate bioactivity data, 2018. https://www.elsevier. com/solutions/reaxys/who-we-serve/pharma-rd/reaxys-medicinalchemistry (accessed Oct, 2018). (41) Kim, S.; Thiessen, P. A.; Bolton, E. E.; Chen, J.; Fu, G.; Gindulyte, A.; Han, L.; He, J.; He, S.; Shoemaker, B. A.; Wang, J.; Yu, B.; Zhang, J.; Bryant, S. H. PubChem substance and compound databases. Nucleic Acids Res. 2016, 44, D1202−D1213. (42) Kim, S.; Thiessen, P. A.; Cheng, T.; Yu, B.; Shoemaker, B. A.; Wang, J.; Bolton, E. V.; Wang, Y.; Bryant, S. H. Literature information in PubChem: Associations between PubChem records and scientific articles. J. Cheminf. 2016, 8, 32. (43) Gaulton, A.; Hersey, A.; Nowotka, M.; Bento, A. P.; Chambers, J.; Mendez, D.; Mutowo, P.; Atkinson, F.; Bellis, L. J.; Cibrián-Uhalte, E.; Davies, M.; Dedman, N.; Karlsson, A.; Magariños, M. P.; Overington, J. P.; Papadatos, G.; Smit, I.; Leach, A. R. The ChEMBL database in 2017. Nucleic Acids Res. 2016, 45, D945−D954. (44) Lipinski, C. A.; Litterman, N. K.; Southan, C.; Williams, A. J.; Clark, A. M.; Ekins, S. Parallel worlds of public and commercial bioactive chemistry data. J. Med. Chem. 2014, 58, 2068−2076. (45) Southan, C.; Várkonyi, P.; Muresan, S. Quantitative assessment of the expanding complementarity between public and commercial databases of bioactive compounds. J. Cheminf. 2009, 1, 10. (46) Tomaszewski, R. Unpublished work, 2017.
94
DOI: 10.1021/acsomega.8b02201 ACS Omega 2019, 4, 86−94
