henhaobio
Sign in

Journal of Proteome ResearchBioinformaticsPeer-reviewed journal article

Cytoscape stringApp: Network analysis and visualization of proteomics data

Nadezhda T. DonchevaiD, John H. MorrisiD, Jan GorodkiniD, Lars Juhl JenseniD

PDFDOIPreprintCode
In-depth full-text digest available · Jump to digest ↓

Abstract

Protein networks have become a popular tool for analyzing and visualizing the often long lists of proteins or genes obtained from proteomics and other high-throughput technologies. One of the most popular sources of such networks is the STRING database, which provides protein networks for more than 2000 organisms, including both physical interactions from experimental data and functional associations from curated pathways, automatic text mining, and prediction methods. However, its web interface is mainly intended for inspection of small networks and their underlying evidence. The Cytoscape software, on the other hand, is much better suited for working with large networks and offers greater flexibility in terms of network analysis, import and visualization of additional data. To include both resources in the same workflow, we created stringApp, a Cytoscape app that makes it easy to import STRING networks into Cytoscape, retains the appearance and many of the features of STRING, and integrates data from associated databases. Here, we introduce many of the stringApp features and show how they can be used to carry out complex network analysis and visualization tasks on a typical proteomics dataset, all through the Cytoscape user interface. stringApp is freely available from the Cytoscape app store: http://apps.cytoscape.org/apps/stringapp.

In-depth summary

Claude Opus 5.5 · xhigh effort · 8 Nov 2026 · from the full text of version 2 (29 Oct 2018) · CC BY 4.0

stringApp, a Cytoscape app, imports STRING and STITCH networks with disease, tissue, localization and drug-target annotations, enabling GUI-based enrichment, clustering and visualization of proteomics data.

The STRING web interface suits only small networks and offers limited analysis flexibility, while Cytoscape handles large networks but has no built-in interaction database. stringApp links the two, bringing STRING queries, styling and associated annotations into Cytoscape so users can combine STRING networks with their own data and other Cytoscape apps.

Key results

  1. ShownstringApp builds STRING or STITCH networks from protein lists, compound lists, disease names or PubMed queries. It attaches tissue, localization, drug-target and evidence-channel attributes as Cytoscape node and edge data.

    Data come from STRING v10.5 (networks for more than 2000 organisms) and STITCH 5, plus DISEASES, TISSUES, COMPARTMENTS and Pharos. Edges carry an overall confidence and per-channel subscores. Lowering the cutoff re-queries the server. The default enrichment threshold is 0.05. · Data sources used by stringApp; Presentation of the stringApp; Scope of the stringApp

  2. ShownIn an ovarian cancer phosphoproteomics use case, nearly all regulated proteins mapped to STRING. Markov clustering in clusterMaker2 then isolated a cluster holding CDK7, POLR2A and splicing proteins, which matches the original study's findings.

    541 unique proteins gave a STRING 10.5 network of 537 nodes and 3027 edges at confidence 0.4. Four proteins were unmapped and 78 singletons were removed. MCL (inflation 4.0, confidence-weighted) kept 1058 within-cluster interactions; the largest cluster had 62 proteins. · Use case: Analysis of an MS-based…; Network retrieval and data import; Using clustering to improve visualization; Functional enrichment analysis

  3. ShownBuilt-in enrichment with Jaccard-based redundancy filtering (Hobohm 1) shortened the largest cluster's term list and pointed to splicing biology. This fits the original report that peptides from splice-variant proteins were overrepresented.

    At FDR 5%, 129 significant terms spanned six categories (GO BP/MF/CC, KEGG, PFAM, InterPro). A redundancy cutoff of 0.5 reduced these to 38. Top terms were GO mRNA processing (39 of 62 proteins) and KEGG Spliceosome (20 of 62), drawn as split donut charts. · Implementation of the app; Functional enrichment analysis

  4. ShownAutomatically retrieved Pharos annotations flagged kinase drug targets, and a separate STRING disease query merged into the study network marked known ovarian cancer genes, without any user-supplied annotation data.

    The cluster held 3 of the 22 kinases in the full network (CDK7, CDK12, CDK13). A DISEASES query for the top 500 epithelial ovarian cancer genes (scores 0.69 to 2.66 on a 0 to 5 scale), filtered to scores above 1.0, gave 222 genes; 13 proteins in the clustered network figure were shown as disease-associated. · Annotation of disease-associated proteins and…; Figure legends

  5. ClaimedNetwork expansion and PubMed-based network retrieval are each scored by a single SQL statement inside a PostgreSQL mirror of STRING and STITCH, so the data never have to be loaded into memory.

    Expansion compares summed confidence to the selected nodes with database-wide connectivity, weighted by selectivity α (default 0.5). PubMed ranking compares mentions in the matching abstracts with corpus-wide mentions, using β fixed at 0.4. Denominators are precomputed. No runtime benchmark is reported. · Algorithms implemented at the database level; Network expansion; PubMed query

Methods, data and reproducibility

A Java app on the Cytoscape 3.6 App API that queries STRING, STITCH, DISEASES and a PostgreSQL mirror through REST APIs. It applies a STRING-like visual style and runs enrichment with redundancy filtering. It is demonstrated on an ovarian cancer phosphoproteomics dataset with clusterMaker2 MCL clustering.

  • Proteins with significantly regulated phosphosites from Francavilla et al. 2017 ovarian cancer phosphoproteomics (Suppl. Table 3; EOC vs OSE and FTE primary cells) · human · n = 541 unique proteins
  • STRING v10.5 protein association network retrieved by protein query for the study proteins · human · n = 537 nodes, 3027 edges
  • DISEASES database gene list for epithelial ovarian cancer retrieved via STRING: disease query · human · n = top 500 candidates; 222 with score above 1.0
  • Cytoscape session with all use-case analyses (the modified input table is Supporting Information 2) · 10.6084/m9.figshare.7258235
Methods & tools
stringApp 1.3.2, Cytoscape 3.6.1, STRING v10.5, STITCH 5, DISEASES, TISSUES, COMPARTMENTS, Pharos, clusterMaker2 1.2.1 (Markov clustering, MCL), EnhancedGraphics 1.2.0, PostgreSQL, cyREST with RCy3 / py2cytoscape
Software & environment
Cytoscape 3.6.1, stringApp 1.3.2, clusterMaker2 1.2.1, EnhancedGraphics 1.2.0. The app is written in Java on the Cytoscape 3.6 App API. Analyses were run on 12 April 2018 against STRING 10.5.
Reproducibility gaps
The use-case session is on figshare, cited by DOI only (see methods.data). No source repository is linked, only the app store. Results rely on live, weekly updated services, so reruns may differ from the April 2018 analysis.
Limitations (5)
  • Tryptic peptides often map to protein groups rather than single proteins. How to show this ambiguity in a protein network (one representative, all members, or fused nodes) is unresolved and stringApp does not address it. · stated by the authors
  • A protein can carry both up- and down-regulated modification sites, but each protein is a single node. Showing every site (for example, as donut plots) overloads large networks. · stated by the authors
  • Large, dense proteomics networks are hard to visualize in ways that reveal patterns, and a single color gradient cannot show multi-condition or time-course comparisons. · stated by the authors
  • The use case is a qualitative re-analysis that recovers findings already reported by the original study. There is no benchmark of runtime or scalability, and no comparison with other network tools. · noted in this summary
  • Network expansion was not used in the use case. Its selectivity default (0.5) and the PubMed β (0.4) are justified by rationale or prior work rather than evaluated here. · noted in this summary

Read this if you need to build, annotate, cluster or run enrichment on STRING protein networks from proteomics or other gene lists in Cytoscape, through the GUI or scripted via cyREST.

Citations
2,319
FWCI
43.63
Field percentile
top 1%

Reproducibility ledger

Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.

Sign in to file a reproducibility report.

Related papers

Similarity Map
SelectedSame fieldCross-field
THISAnderson '25Lima '21Rudolph '19Armanious '20Heberle '17Auer '22
64% semantic matchSystems Biology
Oliver Faulkner Anderson, Altaf Barelvi, Aden O'Brien+3 more·PLoS ONE·Sep 5, 2025