- Open Access
CSgator: an integrated web platform for compound set analysis
Journal of Cheminformaticsvolume 11, Article number: 17 (2019)
Drug discovery typically involves investigation of a set of compounds (e.g. drug screening hits) in terms of target, disease, and bioactivity. CSgator is a comprehensive analytic tool for set-wise interpretation of compounds. It has two unique analytic features of Compound Set Enrichment Analysis (CSEA) and Compound Cluster Analysis (CCA), which allows batch analysis of compound set in terms of (i) target, (ii) bioactivity, (iii) disease, and (iv) structure. CSEA and CCA present enriched profiles of targets and bioactivities in a compound set, which leads to novel insights on underlying drug mode-of-action, and potential targets. Notably, we propose a novel concept of ‘Hit Enriched Assays”, i.e. bioassays of which hits are enriched among a given set of compounds. As an example, we show its utility in revealing drug mode-of-action or identifying hidden targets for anti-lymphangiogenesis screening hits. CSgator is available at http://csgator.ewha.ac.kr, and most analytic results are downloadable.
During the early phase of drug discovery, it is common to identify multiple hit compounds by high-throughput screening (HTS) [1, 2]. It is critical to survey their known targets, activities, and disease indications to avoid potential toxicity or side-effects, to understand structure-activity relations (SAR), and to direct medicinal chemistry for lead generation. Active exploitation of polypharmacology (e.g. dual inhibitors) or drug combination has also been considered as a viable strategy to overcome drug resistance or tumor heterogeneity in cancer therapy . Although scientists have access to many chemogenomic databases, they are not comprehensive enough individually, nor suitable or convenient for batch analyses of a compound set [4–8]. Recent explosion of bioassay datasets (e.g. PubChem and ChEMBL [9, 10]) made rich information available on diverse aspects of bioactivities, but such data have been used only limitedly in drug discovery. Several integrated compound-target DBs are available, but are limited in analytic functions [11, 12]. There were several works on predictive analyses based on bioactivity profiles or fingerprints, most of which did not fully exploited bioactivity data available [13, 14], or were difficult to use for researchers without programming skills .
Here, we developed CSgator (Compound Set navigator), a web platform that provides a comprehensive interpretation of compound set. It is equipped with unique analytic features of Compound Set Enrichment Analysis (CSEA) and Compound Cluster Analysis (CCA). Particularly, we provide unique analytic functions such as HEA analysis (Hit Enrichment Analysis) that provide novel insights or clues on drug mode-of-action, or underlying targets of phenotypic screening hits (e.g. lymphangiogenesis) as described in the following sections with an example case.
Materials and methods
Standardization of compound IDs and gene names
In order to avoid redundancy, CSgator amassed a consolidated set of compounds from public chemical database such as PubChem, ChEMBL, ChEBI, and DrugBank [5, 7, 10, 16]. We then merged different isotopic, (un)charged, and (de)protonated forms of the same molecule into a single compound ID. For example, lovastatin, a HMG-CoA reductase inhibitor falls into 62 PubChem CIDs, all of which would show essentially the same or highly similar biological activity. All the compounds were mapped to a unified compound ID based on IUPAC InChIKey (IUPAC International Chemical Identifier Key) using Open Babel v.2.3  by converting SMILES or MOL format to InChIKey strings as well as by manual mapping of compound names where necessary. Gene IDs were standardized using the gene names given by UniProtKB and NCBI Gene [18, 19].
Collection of compound-target interaction data
We collected compound-target interaction data from 15 public databases: CTD, DCDB, DrugBank, MATADOR, TTD, BindingDB, ChEMBL, KiDB, KEGG Drug, PharmGKB, IUPHAR, Binding MOAD, DGIdb, GLASS, STITCH [4, 6,7,8, 10, 20–29]. After ID standardization of compounds and genes, a total of > 3 mil. compound-target interactions are collected (Table 1).
Classification of targets and diseases
Compounds and targets were classified by four different annotations: (I) Protein family classes by ChEMBL version 21, (II) Gene Ontology (GO) terms on Biological Process (BP) , (III) Disease Ontology (DO) terms that cross reference with MeSH (Medical Subject Headings), ICD (International Classification of Diseases), NCI’s thesaurus, SNOMED (Systemized Nomenclature of Medicine) and OMIM , and (IV) MeSH Disease term provided by NLM (U.S. National Library of Medicine) .
Bioassay data from PubChem Bioassay and ChEMBL
Bioassay data include information for diverse aspects of compound bioactivities. We collected over 1.2 mil. bioassay dataset for > 2 mil. compounds from PubChem Bioassay and ChEMBL [9, 10]. Some of the bioassay dataset were not in a standardized format and required further processing such as ordering compounds by activity, assignment of hit/non-hit compounds, and target ID standardization for targeted bioassays. We assigned compounds as hit by applying one of the three criteria. First, PubChem Bioassay and ChEMBL provide active/inactive information for ~ 22% of the total assays (~ 270,000 bioassays), and accordingly, we took the information to assign hit or non-hit compounds. For the remaining bioassays without active/inactive annotation, the cut-off of Z score ≥ 2 or the top 1% were applied as the second and the third criteria, and took the union of the resulting compound sets as hits. Because only a small fraction of the assays were annotated to a specific target, we also performed a manual curation to assign bioassays to a specific target whenever target information is available in the assay title or description. As a result, ~ 10.3% of the total assays were assigned to a specific target.
Generation of structural properties
We calculated structural and physicochemical properties of all the compounds, which can be exploited for characterization or filtering of a compound set using Open Babel toolbox . The physicochemical properties were calculated such as molecular weight, FP2 fingerprint, logP (Partition coefficient), topological polar surface area (TPSA), and hydrogen bond donor and acceptor. Additionally, we generate predictors for lead-likeness, i.e. Lipinski’s the rule, and QED (quantitative estimation of drug-likeness) by Gregory Gerebtzoff (Roche, Switerland) implemented in Silico-it package .
Utility and discussion
CSgator includes information on ~ 90 million compounds after merging redundant entries, > 6 million compound-target relations from 15 public databases (Table 1), ~ 1.6 million compound-disease associations, and > 230 million bioactivity points collected from > 1.2 million bioassay data set. Whenever available, compounds and targets were annotated by protein family, functional annotation by Gene Ontology , and disease categories by Disease Ontology and MeSH [31, 32]. As described in the following sections, these annotations are crucial to interpret the characteristics of input compound set, and provide novel clues on drug mode-of-action, and will be expanded as more information accumulate. Data sources and current statistics are listed in Table 2. These data may be available elsewhere, but CSgator is unique for its comprehensiveness, clean mapping between different resources, and full data accessibility.
The analytic workflow of CSgator consists of three steps of (i) generation of input compound set, (ii) tabular listing of annotations for the input compound set, and (iii) compound set analysis step as depicted in Fig. 1. First, input compound set can be generated in three different ways: (a) by Compound ID Search using SMILES, InChI, InChIKey, CAS Registry Number, and other IDs including PubChem, ChEMBL, ChEBI, and DrugBank, (b) by Compound Structure Search for compounds with specific scaffolds or by structural similarity, and (c) by Compound Set Selection, where the precompiled compound set is selected. Precompiled compound sets were built in various ways, e.g. by target or target family, approval status by FDA and other countries, and disease indication. Notably, users can also freely generate a new compound set by applying Set Operator to precompiled or input compound sets, and by filtering compounds based on physicochemical properties. Second, CSgator internally gathers all the annotations of input compounds that are grouped into four categories: (i) target, (ii) bioassay, (iii) disease, and (iv) structure. All the annotations are listed and downloadable in a tabular format. Third, user can investigate collective information of a compound set, which is not available in other related databases. The two unique analyses in CSgator are CSEA (Compound Set Enrichment Analysis) and CCA (Compound Cluster Analysis), which will be further explained in the following sections.
Compound Set Enrichment Analysis (CSEA)
Similarly to Gene Set Enrichment Analysis (GSEA ), Compound Set Enrichment Analysis (CSEA) refers to investigating enriched annotations for a compound set. Varin et al.  applied CSEA to identify active scaffolds enriched in primary screening data. We extend CSEA even further to annotations on target, disease, and bioassay hits. Particularly, we propose a novel concept of Hit Enriched Assays (HEAs) as bioassays of which hits are enriched among the compound set of interest. Since bioassays generally have intended targets and biological processes, HEAs can provide non-obvious links to the underlying targets and drug mode-of-actions enriched in the input compound set such as phenotypic screening hits. Similarly, it also shows enriched targets or diseases in a tree format, i.e. Target Enrichment Tree (TET), and Disease Enrichment Tree (DET). The degree of enrichment, or Enrichment Score (ES) is calculated as log likelihood ratio (LLR) for HEAs, and odds ratio for TET and DET.
where Q is the input or query compounds, and QC is the compounds that do not belong to Q. H is the compounds of interest, e.g. hit compounds for HEA, ligands for a target or target family for TET, and compounds related to a disease for DET analysis. In calculating ES for TET (or DET), we assume compounds not in the query QC do NOT interact with the target or target family (or disease) although more interactions may exist, but not yet discovered in any test. These missing information may skew the results of TET (or DET), which should be cautiously interpreted. Later, we show an example case of CSEA in interpreting anti-lymphangiogenesis screening hits in the ‘Case Study’ section below.
Compound Cluster Analysis (CCA)
Structurally similar compounds tend to share the same or structurally similar targets . With the purpose to investigate this aspect, CSgator first generates Compound Clusters (CCs) of structurally similar subgroups by k-means clustering. It then shows Compound Cluster Network (CC-Network), showing connections among the compound clusters with target family or disease classes. Similarly to CSEA, the degree of enrichment for each CC is also calculated as odd ratio, where R represent the compounds of each cluster (CC), and RC is the all other compounds in the database. Therefore, CC-Network provides information on how a structurally similar cluster of compounds (CC) would be significantly associated to a specific target family or disease class compared to all other compounds as background.
A case study on interpreting phenotypic screening hits
Here, we show the utility of the two main analytic functions in CSgator using a case study in interpreting phenotypic screening hits. We took the high-content phenotype-based assay data for screening inhibitors of lymphangiogenesis . Schulz et al. screened FDA approved 1280 drugs (Library of Pharmacologically Active Compounds or LOPAC library from Sigma), resulting in identifying 31 hits (hit rate of 2.4%). The 31 hits were mapped to 40 unique compound IDs in CSgator. But this screening dataset alone does not provide information on the underlying targets or drug mode-of-action. With the 40 compounds as an input set, we performed CSEA and CCA analyses implemented in CSgator as described in the following section.
CSEA (Compound Set Enrichment Analysis)
CSEA investigates enriched annotations in terms of target, disease, and bioactivities. In HEA analysis, CSgator listed 146 bioassays, where the 40 input compounds were significantly enriched as hits. We took the list of HEAs that have explicit information on their intended targets with high enrichment score (ES > 5) as listed in Table 3. The targets of the top ranked HEAs include many genes that were known to be involved in lymphangiogenesis. The top ranking HEA (ES = 9.75) screened for RGS4 (Regulator of G-protein signaling 4) inhibitors. Indeed, RGS4 plays a key role in regulating tubulogenesis including lymphangiogenesis by antagonizing MAPK and VEGF signaling [38, 39]. The third and fifth HEA targeted mTOR (ES = 7.12), which generally known to control lymphangiogenesis [40, 41]. Thrombopoietin (TPO) is the regulator of thrombocyte production, and recent studies provide evidence for the critical role of the thrombocytes in lymphangiogenesis in human malignant tumors [42, 43]. A bioassay targeting TPO was ranked at the top 7th with ES = 6.21. Vascular endothelial growth factor D (VEGF-D) has been implicated in the key role of lymphangiogenesis. TNF-α induces AP-1 binding to the VEGF-D promoter, and increase VEGF-D expression through TNF-α/ERK1/2/AP-1 pathway, which promotes lymphangiogenesis and lymphatic metastasis [44, 45]. The 10th HEA (ES = 5.85) targeted AP1 signaling. In summary, five out of the top 10 HEAs provided direct links to the known genes associated to lymphangiogenesis. Accordingly, other targets of high ranking HEAs may be also involved in lymphangiogenesis, such as GMNN, ATAD5, ATXN2, and FEN1 (Table 3).
Similarly, we performed Target Enrichment Tree (TET) analysis to get useful clues to underlying targets of phenotypic screening assays. CSgator listed target protein families prioritized by enrichment score (Table 4). The top ranked target family was calcium-activated chloride channel family (ES = 5.90), of which key role was reported in lymph node remodeling by induction of lymphangiogenesis . Among the 40 input compounds, only 2 compounds are known to interact with the targets of the family members in our dataset. It demonstrates the current lack of enough compound-target data even after the integration of 15 publicly available datasets. Therefore, the utility of TET or DET analysis may be limited at the moment compared to HEA analysis. In spite of this limitation, it showed significant enrichment (FDR-adjusted p value = 0.00431). We were able to identify other target families potentially associated to lymphangiogenesis. The families related to cytochrome P450 were found frequently within the top 10 ranks (five out of the ten families). It may be associated that oxygen released by oxidoreduction in lymph tissue causes expansion of lymphatic vessels . If we use a larger input set and collect more compound-target dataset, TET analysis may become more useful with better statistical power.
CCA (Compound Cluster Analysis)
Certain properties of a compound set may be evident only in structurally similar subgroups. CCA allows identification of enriched features in structurally similar clusters of compounds. In CSgator, we obtained three compound clusters (CC #1–#3) in the 40 input compounds by setting the number of clusters, k = 3. Then, a network of CCs and disease classes is generated (Fig. 2). This network showed the distribution of their original indications, and several notable connections were observed. CC #2 was linked to several diseases including viral infectious disease. There are several studies that herpes virus-triggered immune response drives lymphangiogenesis [39, 48, 49]. Both CC #2 and #3 were strongly connected to cancer, which may be expected because inhibition of lymphangiogenesis has emerged as a promising strategy for cancer therapy [47, 50].
CSgator is a highly comprehensive and integrated analytic system for compound set analysis in terms of targets, bioactivity profiles, structural properties, and disease indications. Such information is crucial to interpret a set of compounds such as high-throughput screening hits, avoid potential side effects or toxicity, and investigate polypharmacology profiles for drug discovery and development. It provides unique functions such as CSEA and CCA, which are not available in other similar tools and databases. It showed that CSgator can give novel clues on drug mode-of-action and the underlying targets for phenotypic screening hits, as shown in the example case of interpreting the anti-lymphangiogenesis screening hits.
Compound Set Enrichment Analysis
Compound Cluster Analysis
Hit Enriched Assay
log likelihood ratio
International Chemical Identifier
Comparative Toxicogenomics Database
Medical Subject Headings
National Institutes of Health
NCATS Chemical Genomics Center
quantitative estimation of drug-likeness
Chemical Abstracts Service
quantitative high-throughput screening
Bleicher KH, Böhm H-J, Müller K, Alanine AI (2003) Hit and lead generation: beyond high-throughput screening. Nat Rev Drug Discov 2:369–378. https://doi.org/10.1038/nrd1086
Macarron R, Banks MN, Bojanic D et al (2011) Impact of high-throughput screening in biomedical research. Nat Rev Drug Discov 10:188–195. https://doi.org/10.1038/nrd3368
Antolin A, Workman P, Mestres J, Al-Lazikani B (2017) Polypharmacology in precision oncology: current applications and future prospects. Curr Pharm Des 22 (46):6935–6945
Chen X, Ji ZL, Chen YZ (2002) TTD: therapeutic target database. Nucleic Acids Res 30:412–415
Degtyarenko K, de Matos P, Ennis M et al (2008) ChEBI: a database and ontology for chemical entities of biological interest. Nucleic Acids Res 36:D344–D350. https://doi.org/10.1093/nar/gkm791
Gunther S, Kuhn M, Dunkel M et al (2007) SuperTarget and Matador: resources for exploring drug-target relationships. Nucleic Acids Res 36:D919–D922. https://doi.org/10.1093/nar/gkm862
Law V, Knox C, Djoumbou Y et al (2014) DrugBank 4.0: shedding new light on drug metabolism. Nucleic Acids Res 42:D1091–D1097. https://doi.org/10.1093/nar/gkt1068
Sharman JL, Benson HE, Pawson AJ et al (2013) IUPHAR-DB: updated database content and new features. Nucleic Acids Res 41:D1083–D1088. https://doi.org/10.1093/nar/gks960
Wang Y, Suzek T, Zhang J et al (2014) PubChem BioAssay: 2014 update. Nucleic Acids Res 42:D1075–D1082. https://doi.org/10.1093/nar/gkt978
Bento AP, Gaulton A, Hersey A et al (2014) The ChEMBL bioactivity database: an update. Nucleic Acids Res 42:D1083–D1090. https://doi.org/10.1093/nar/gkt1031
Irwin JJ, Sterling T, Mysinger MM et al (2012) ZINC: a free tool to discover chemistry for biology. J Chem Inf Model 52:1757–1768. https://doi.org/10.1021/ci3001277
Roider HG, Pavlova N, Kirov I et al (2014) Drug2Gene: an exhaustive resource to explore effectively the drug-target relation network. BMC Bioinformatics 15:68. https://doi.org/10.1186/1471-2105-15-68
Cheng T, Li Q, Wang Y, Bryant SH (2011) Identifying compound-target associations by combining bioactivity profile similarity search and public databases mining. J Chem Inf Model 51:2440–2448. https://doi.org/10.1021/ci200192v
Helal KY, Maciejewski M, Gregori-Puigjané E et al (2016) Public domain HTS fingerprints: design and evaluation of compound bioactivity profiles from PubChem’s bioassay repository. J Chem Inf Model 56:390–398. https://doi.org/10.1021/acs.jcim.5b00498
William T, Backman H, Girke T (2016) bioassayR: cross-target analysis of small molecule bioactivity. J Chem Inf Model 9:99. https://doi.org/10.1021/acs.jcim.6b00109
Kim S, Thiessen PA, Bolton EE et al (2016) PubChem substance and compound databases. Nucleic Acids Res 44:D1202–D1213. https://doi.org/10.1093/nar/gkv951
O’Boyle NM, Banck M, James CA et al (2011) Open Babel: an open chemical toolbox. J Cheminform 3:33. https://doi.org/10.1186/1758-2946-3-33
Magrane M, UniProt Consortium (2011) UniProt Knowledgebase: a hub of integrated protein data. Database (Oxford) 2011:bar009. https://doi.org/10.1093/database/bar009
Maglott D, Ostell J, Pruitt KD, Tatusova T (2011) Entrez gene: gene-centered information at NCBI. Nucleic Acids Res 39:D52–D57. https://doi.org/10.1093/nar/gkq1237
Liu Y, Wei Q, Yu G et al (2014) DCDB 2.0: a major update of the drug combination database. Database (Oxford) 2014:bau124. https://doi.org/10.1093/database/bau124
Davis AP, Grondin CJ, Johnson RJ et al (2017) The Comparative Toxicogenomics Database: update 2017. Nucleic Acids Res 45:D972–D978. https://doi.org/10.1093/nar/gkw838
Gilson MK, Liu T, Baitaluk M et al (2016) BindingDB in 2015: a public database for medicinal chemistry, computational chemistry and systems pharmacology. Nucleic Acids Res 44:D1045–D1053. https://doi.org/10.1093/nar/gkv1072
Benson ML, Smith RD, Khazanov NA et al (2008) Binding MOAD, a high-quality protein-ligand database. Nucleic Acids Res 36:D674–D678. https://doi.org/10.1093/nar/gkm911
Wagner AH, Coffman AC, Ainscough BJ et al (2016) DGIdb 2.0: mining clinically relevant drug-gene interactions. Nucleic Acids Res 44:D1036–D1044. https://doi.org/10.1093/nar/gkv1165
Chan WKB, Zhang H, Yang J et al (2015) GLASS: a comprehensive database for experimentally validated GPCR-ligand associations. Bioinformatics 31:3035–3042. https://doi.org/10.1093/bioinformatics/btv302
Szklarczyk D, Santos A, von Mering C et al (2016) STITCH 5: augmenting protein–chemical interaction networks with tissue and affinity data. Nucleic Acids Res 44:D380–D384. https://doi.org/10.1093/nar/gkv1277
Thorn CF, Klein TE, Altman RB (2013) PharmGKB: the pharmacogenomics knowledge base. Methods Mol Biol 1015:311–320. https://doi.org/10.1007/978-1-62703-435-7_20
Kanehisa M, Furumichi M, Tanabe M et al (2017) KEGG: new perspectives on genomes, pathways, diseases and drugs. Nucleic Acids Res 45:D353–D361. https://doi.org/10.1093/nar/gkw1092
Roth BL, Lopez E, Patel S, Kroeze WK (2000) The multiplicity of serotonin receptors: uselessly diverse molecules or an embarrassment of riches? The Neuroscientist 6:252–262. https://doi.org/10.1177/107385840000600408
Gene Ontology Consortium (2015) Gene ontology consortium: going forward. Nucleic Acids Res 43:D1049–D1056. https://doi.org/10.1093/nar/gku1179
Schriml LM, Arze C, Nadendla S et al (2012) Disease ontology: a backbone for disease semantic integration. Nucleic Acids Res 40:D940–D946. https://doi.org/10.1093/nar/gkr972
Erwin PJ, Perkins WJ (2007) Medline: a guide to effective searching in PubMed & other interfaces, 2nd Edition. Anesthesiology 107:360–361. https://doi.org/10.1097/01.anes.0000271865.33903.be
Bickerton GR, Paolini GV, Besnard J et al (2012) Quantifying the chemical beauty of drugs. Nat Chem 4:90–98. https://doi.org/10.1038/nchem.1243
Subramanian A, Tamayo P, Mootha VK et al (2005) Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci USA 102:15545–15550. https://doi.org/10.1073/pnas.0506580102
Varin T, Gubler H, Parker CN, Zhang J-H, Raman P, Ertl P, Schuffenhauer A (2010) Compound set enrichment: a novel approach to analysis of primary HTS data. J Chem Inf Model 50 (12):2067–2078
Martin YC, Kofron JL, Traphagen LM (2002) Do structurally similar molecules have similar biological activity? J Med Chem. https://doi.org/10.1021/JM020155C
Schulz MMP, Reisen F, Zgraggen S et al (2012) Phenotype-based high-content chemical library screening identifies statins as inhibitors of in vivo lymphangiogenesis. Proc Natl Acad Sci USA 109:E2665–E2674. https://doi.org/10.1073/pnas.1206036109
Albig AR, Schiemann WP (2005) Identification and characterization of regulator of G protein signaling 4 (RGS4) as a novel inhibitor of tubulogenesis: RGS4 inhibits mitogen-activated protein kinases and vascular endothelial growth factor signaling. Mol Biol Cell 16:609–625. https://doi.org/10.1091/mbc.e04-06-0479
Aguilar B, Choi I, Choi D et al (2012) Lymphatic reprogramming by Kaposi sarcoma herpes virus promotes the oncogenic activity of the virus-encoded G-protein-coupled receptor. Cancer Res 72:5833–5842. https://doi.org/10.1158/0008-5472.CAN-12-1229
Chen H, Guan R, Lei Y et al (2015) Lymphangiogenesis in gastric cancer regulated through Akt/mTOR-VEGF-C/VEGF-D axis. BMC Cancer 15:103. https://doi.org/10.1186/s12885-015-1109-0
Ekshyyan O, Moore-Medlin TN, Raley MC et al (2013) Anti-lymphangiogenic properties of mTOR inhibitors in head and neck squamous cell carcinoma experimental models. BMC Cancer 13:320. https://doi.org/10.1186/1471-2407-13-320
Bertozzi CC, Hess PR, Kahn ML (2010) Platelets: covert regulators of lymphatic development. Arterioscler Thromb Vasc Biol 30:2368–2371. https://doi.org/10.1161/ATVBAHA.110.217281
Schoppmann SF, Alidzanovic L, Schultheis A et al (2013) Thrombocytes correlate with lymphangiogenesis in human esophageal cancer and mediate growth of lymphatic endothelial cells in vitro. PLoS ONE 8:e66941. https://doi.org/10.1371/journal.pone.0066941
Hong H, Jiang L, Lin Y et al (2016) TNF-alpha promotes lymphangiogenesis and lymphatic metastasis of gallbladder cancer through the ERK1/2/AP-1/VEGF-D pathway. BMC Cancer 16:240. https://doi.org/10.1186/s12885-016-2259-4
Lin W, Jiang L, Chen Y et al (2012) Vascular endothelial growth factor-D promotes growth, lymphangiogenesis and lymphatic metastasis in gallbladder cancer. Cancer Lett 314:127–136. https://doi.org/10.1016/j.canlet.2011.09.004
Jordan-Williams KL, Ramanujam N, Farr AG, Ruddell A (2016) The lymphatic endothelial mCLCA1 antibody induces proliferation and growth of lymph node lymphatic sinuses. PLoS ONE 11:e0156079. https://doi.org/10.1371/journal.pone.0156079
Stacker SA, Achen MG (2008) From anti-angiogenesis to anti-lymphangiogenesis: emerging trends in cancer therapy. Lymphat Res Biol 6:165–172. https://doi.org/10.1089/lrb.2008.1015
Sessa R, Chen L (2017) Lymphangiogenesis: a new player in herpes simplex virus 1-triggered T-cell response. Immunol Cell Biol 95:5–6. https://doi.org/10.1038/icb.2016.108
Wuest TR, Carr DJJ (2010) VEGF-A expression by HSV-1-infected cells drives corneal lymphangiogenesis. J Exp Med 207:101–115. https://doi.org/10.1084/jem.20091385
Nisato RE, Tille J-C, Pepper MS (2003) Lymphangiogenesis and tumor metastasis. Thromb Haemost 90:591–597. https://doi.org/10.1160/TH03-04-0206
WK designed the study, and SP, YK, HJ, SJ, HL collected data and performed the analysis, and SP implemented the web site. SP and WK wrote the manuscript. All authors read and approved the final manuscript.
We thank Seungmin Kang for helpful discussions and comments.
The authors declare that they have no competing interests.
Availability and data and materials
The web platform can be accessible at http://csgator.ewha.ac.kr.
This work was funded by Grants from National Research Foundation of Korea (NRF-2017R1A2B4007855, and NRF-2017M3C9A5028690) and from Institution for Information and communications Technology Promotion(IITP) (No. 2016-0-00289).
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.