TL;DR
PhosphoSitePlus (PSP) has been a cornerstone resource for post-translational modification (PTM) data for 15 years. This paper details significant expansions to PSP, integrating disease variants and protein isoforms, enabling researchers to explore the complex interplay between genetic mutations, PTMs, and cellular function, particularly in disease contexts.
What Problem Does This Paper Address?
Post-translational modifications (PTMs) are crucial for regulating cellular processes, but their dynamic nature and complex interactions with other cellular events make them challenging to study. Furthermore, genetic variations (mutations) can directly impact PTM sites, potentially altering signaling pathways and leading to disease. Understanding these PTMs, their regulation, and how they are affected by genetic mutations is vital for deciphering disease mechanisms and identifying therapeutic targets. Existing resources often focus on individual aspects (PTMs, mutations, or isoforms) in isolation. This paper addresses the need for a unified resource that integrates these critical layers of biological information, providing a more holistic view of protein function and dysfunction.
Methods
The paper highlights several key methodological advancements and data integration strategies:
- Data Curation: PSP continues to rely on manual curation by expert scientists for PTM data, ensuring accuracy and depth. This includes information on upstream regulation (ligands, enzymes) and downstream consequences of PTMs.
- Disease Variant Integration: Germline mutations associated with inherited diseases and somatic cancer mutations are now integrated into PSP. This involves mapping these variants to proteins and PTM sites.
- Isoform Handling: A new algorithm, Mirage, is introduced for splice-aware multiple sequence alignment. This is crucial for accurately aligning protein isoforms, which can differ significantly due to alternative splicing, and is essential for comprehensive PTM and variant annotation across these isoforms.
- Interactive Visualization: The introduction of "lollipop plots" provides a novel way to visualize PTM sites, disease variants (both germline SNPs and somatic mutations), and quantitative data (HTP/LTP counts, MAFs, tumor frequencies) in a single, interactive interface.
Datasets
- PTM Data: Curated from over 22,000 articles and thousands of mass spectrometry (MS) datasets.
- Disease Variants:
- Germline mutations from ClinVar (July 2018 release), including 3265 proteins with pathogenic or likely pathogenic missense mutations.
- Germline missense mutations from the UniProtKB Humsavar dataset (June 2018 release), comprising over 30,191 disease variants.
- Somatic cancer mutations from 15 published TCGA datasets, covering 4,440 tumors across 15 cancer types.
- Protein Isoforms: Over 30,000 human and mouse UniProtKB/Swiss-Prot isoform sequences are being imported.
- MS2 Data: Quantitative MS2 data sets are being curated for future visualization tools.
Key Insights and Ideas
- The PTM-Mutation Nexus: The integration of disease variants with PTM data is a major step forward. It allows researchers to directly visualize how genetic changes might disrupt PTM sites, potentially altering signaling networks and contributing to disease. This provides a powerful framework for hypothesis generation.
- Isoforms Matter: The paper emphasizes the growing importance of studying protein isoforms, which arise from alternative splicing. These isoforms can have distinct functions and PTM profiles, and accurately aligning them is critical for comprehensive analysis. The new Mirage algorithm is a key enabler for this.
- Visualizing Complexity: The interactive lollipop plots are a significant advancement, transforming complex data into an intuitive visual format. This allows for easier exploration of the relationships between PTMs, mutations, and their associated quantitative data, making it more accessible to a broader range of researchers.
- Data-Driven Discovery: The sheer volume of data curated (over 450,000 PTM sites) and the ongoing expansion highlight the power of large-scale data integration for uncovering biological insights.
Key Results
- Database Growth: PSP now contains over 450,000 unique PTM sites from over 22,000 articles and thousands of MS datasets, a 30% increase since 2015.
- Increased Coverage: Significant increases in curated sites for sumoylation (10-fold), monomethylation (3-fold), and ubiquitylation (1.5-fold) were observed.
- Isoform Expansion: The database is actively incorporating over 30,000 human and mouse isoforms.
- Disease Variant Overlap: Analysis revealed that 420 genetic variants overlap with 402 PTM sites on 276 proteins.
- Performance Improvement: Query optimization has improved search speeds by a factor of 15, with average protein searches now taking less than two seconds.
Limitations and Future Work
- Limitations:
- The paper acknowledges that the selection of somatic hotspot mutations may be biased towards cancer types with larger cohorts.
- While PTMVars are identified, the paper suggests that incorporating scores for the effect of missense mutations on protein stability would be valuable.
- Future Work:
- Quantitative Data Viewer: Development of tools to visualize quantitative MS2 data.
- Regulatory Interaction Visualization: Integration of a web visualization tool for PTM-centric regulatory data and molecular interactions.
- Kinase and Binding Partner Prediction: Integration of a kinase predictor into PSP.
- Continued Data Integration: Ongoing updates to germline and somatic nsSNP data, and potential incorporation of mutations from additional cancer types.
- Improved Alignment: The Mirage algorithm is a significant step towards better isoform alignment, with further improvements anticipated.
Summary updated