Protein Modification and Drug Sensitivity

PMADS workflow

We developed a literature mining pipeline using search terms derived from the dbPTM vocabulary and automated article retrieval via NCBI E-Utilities (EDirect) combined with Biopython. To broaden our corpus, we utilized the NCBI TranslationSet for thesaurus expansion of both PTM and keyword ontologies. Entity recognition was performed using PubTator and bespoke regular expressions to capture multi‑class biomedical entities, underpinned by pretrained natural language processing models (Stanza and CRAFT) for relation extraction. Candidate associations were filtered via manually curated inter‑entity rules, followed by expert review for final validation. To date, the data pipeline has processed 20,310,267 documents. Through rigorous manual curation, over 4,500 post-translational modification (PTM)-drug associations have been identified, among which more than 2,500 are cancer-related.

Workflow of PMADS

In addition to curated literature-derived associations, PMADS includes predicted PTM–drug–disease associations inferred from large-scale public proteomics data. We analyzed 50 PRIDE datasets using MaxQuant for peptide identification and PTM site quantification, retaining only sites with MaxQuant Score > 40, localization probability > 0.8, and Delta score > 40, while excluding potential contaminants and reverse sequences. Differential analysis with Wilcoxon rank-sum tests identified PTM changes with Benjamini–Hochberg–adjusted p < 0.05, alongside changes with unadjusted p < 0.05 and absolute log₂ fold change> 10, the latter classified as low-confidence. The current release contains over 43,800 inferred associations spanning 7,878 proteins, 24,195 PTM types, 39 drugs, and 18 disease types, primarily in cancer.

PTM types in PMADS

PTM types in PMADS

Diseases in PMADS

Disease category

Key features

Applications

Basic Research: Elucidate molecular networks of PTM-regulated drug mechanisms.

Drug Development: Guide targeted drug design by mapping "druggable" PTM hotspots.

Precision Medicine: Identify patient-specific PTM signatures for personalized treatment optimization.

Resource switch

On the right side of the database page title, an icon can be clicked to switch between data sources: Curated and Inferred, with the default setting being "Curated". The "Curated" data records are derived from literature reports in PubMed, while the "Inferred" data records originate from PRIDE data analysis. After switching, all search and autocomplete functions will be adjusted accordingly. Users can switch data sources on any page and search for content of interest.

Fig. 1 Resource switch position

Quick search

Located in the upper-right corner of every page, the quick search interface offers a unified portal with four clickable icons that enable rapid access to distinct categories: PTM sites, proteins, genes, and drugs. Users can initiate searches with minimal input, and the system returns matched entries enriched with PTM types, counts, and contextual annotations.

Fig. 2 Quick search entry

Quick search for protein

Click the protein icon (selected by default) to switch the search box to protein quick search, which supports searching by protein name or Uniprot ID, and will return fuzzy search results.

Fig. 3 Quick search for protein

On the search results page (using VEGFR2 as an example), the search statistics and categorized PTM statistics results page will be displayed (Fig.3). The results page splits the data by protein name, ID, and species, and displays the types and categories of post-translational modifications for different proteins on the right side. P denotes phosphorylation, Me denotes methylation, Ub denotes ubiquitination, and Misc denotes other types. You can view detailed records by clicking on the protein name or different PTMs (Fig. 4).

Fig. 4 Results of quick search for protein

Fig. 5 Results of PTM (phosphorylation of VEGFR2)

Quick search for gene

Gene quick search can be switched by clicking on the gene icon, and supports fuzzy input and auto-complete (e.g., SMAD3).

Fig. 6 Quick search for gene

Fig. 7 Results of quick search for gene

Quick search for drug

Drug quick search can be switched by clicking on the drug icon (e.g., imatinib).

Fig. 8 Quick search for drug

Fig. 9 Quick search results of drug

Quick search for PTM site

The rapid search function of the Site enables the retrieval of proteins and residue positions. Searches can be conducted using either the protein name followed by the amino acid residue and its position (e.g., EGFR-Tyr1068) or the UniProt ID along with the amino acid residue and its position (e.g., P00533-Tyr1068).

Fig. 10 Quick search for PTM site

Fig. 11 Results of quick search of PTM site

Advanced search

Available under the Search page, supporting combinatorial queries across multiple dimensions (Drug, Protein, Disease, Species, PTM type and Site ). The "DEMO" button can automatically populate search examples, and clicking the "SEARCH" button will allow access to the search results.

Fig. 12 Advanced search page

Result of record

The sole means of accessing detailed result data from PMADS is by clicking on the Record ID (e.g., acc0079) displayed on the search results page (Fig. 4).

On the PMADS record page, there is a detailed description of the record, including the following:

Usage of API

This API provides flexible search functions for post-translational modifications of proteins and related data in the PMADS database.

API basic information

Request parameters

Response format

Examples

Citation

PMADS (Protein Modification and Drug Sensitivity). 2025. https://pmads-db.org/

Download

Currently, PMADS supports full database downloads, core data downloads, and annotation data downloads. You can select the desired download method by navigating to the Download page.

Abbreviations in PMADS

Provides amino acid abbreviations and other abbreviations for easy user lookup and use of PMADS.

Amino acid abbreviations

Other abbreviations

Effect & Score

In the detailed information of records in PMADS, Relation, Effect, Effect Info, Note, and Score are key record entries.

Ternary Association: The diagram describing the relation of PTM-Drug-Disease is defined as three patterns:

Pattern 1: PTM changes directly influencing drug efficacy in disease (solid line).

Pattern 2: drugs modulating PTM states that subsequently affect disease outcomes (dashed line).

Pattern 3: drugs simultaneously regulating PTM alterations and disease progression (grey dashed line).

PMADS pattern of Ternary Diagram

Regulatory Class: Describes the regulatory of increased PTM on the efficacy of drugs in diseases.

Effect Info: Provides a detailed description for each record, specifically indicating the influence of PTM at a certain site of a specific protein on the treatment of a particular disease with a specific drug.

Note: Serves as supplementary explanation to the Effect of the record, marking information such as combined medication, unconventional drugs, and whether it is a histone, etc.

Score: The impact score is used by PMADS to quantify the effect of PTM on the efficacy of drugs in diseases. The definition of this indicator is as shown in the following table:

Contact us

If you have any questions, please feel free to contact us. Suggestions and bug reports are welcome.

E-mail: [email protected]

Telephone: +8621-65980233

Address: No. 1239, Siping Road, Yangpu Area, Shanghai, China