PTMFunc DATABASE
PTM Functional Landscape
“Post‐Translational Modification Functional Landscape Database (PTMFunc DATABASE)” presents a new strategy with the introduction of functional PTMs to enlarge the targetable biological space. Taking phosphorylation as an example, > 51000 phosphosites have been reported at human proteomics level. Whereas only 7402 phosphosites were identified with both kinase and regulatory information, occupying less than 2% of the reported phosphosites. The unexplored huge space for the identification of kinase-phosphosite and phosphosite functional assignment just represents the “Dark matter” in phosphorylation research.
PTMFunc database provides a platform for exhaustive information on PTMs, mainly for phosphosites now, and their functional annotations. The database now contains 3335 proteins and 135063 phosphosites from human . Phosphosites are annotated with detailed annotations of their molecular functions, biological processes, and molecular interactions from PhosphoSitePlus (PSP) database. The ranked first three molecular functions for phosphosites include molecular association, activity regulation and intracellular localization. The ranked first three biological processes for phosphosites are cell cycle regulation, transcription induced and signaling pathway regulation. More than half phosphosites are involved in both molecular functions and biological processes. In addition, phosphosites were annotated with disease information from PSP and PTMD databases, and indicated the increasing involvement of aberrant phosphosites in various cancers and neurodegenerative diseases.
Top Menu: The top menu contains six menus.
Main Frame: The main frame organizes pages containing the PTMFunc DATABASE entry list, reference, quick links and visit browse, in the multiple tab panels form. There are two types of functional phosphosite prediction, sequence-based and structure-based, which can be selected in Tools, and users can click DownLoad to view site statistics and download relevant protein sequences In PTMFunc DATABASE, related literature, phosphosite information and protein databases can be quickly accessed through links.
Footer: The Footer helps users to quickly contact the author.
The functional phosphosite data used in this work were obtained from the PhosphoSitePlus (PSP, 2022.7database). Regulatory_sites dataset from PSP contains 18518 PTM loci with reported regulatory functions, including detailed information of molecular functions, biological processes, and intermolecular interactions of the site. Focusing on human protein phosphosites, 9769 regulatory phosphosite (135063 phosphosites in all) in 3335 human proteins were finally collected. In addition, the complementary information of the phophosites, e.g., kinases, and diseases, was supplemented. The numbers of phosphosites and phosphoproteins collected are shown in the table and figures.
| Phosphoprotein | Phosphosite | |
|---|---|---|
| Reported kinases | 2658 | 10697 |
| Reported regulatory | 3054 | 9769 |
| +Reported function | 2838 | 8625 |
| +Reported process | 1884 | 4663 |
| Disease-associated (PSP) | 566 | 1054 |
| Reported disease (PTMD) | 310 | 551 |
| Total | 3335 | 135063 |
In the search page, proteins are categorized according to family information, and by clicking on specific categories, you can browse information about the proteins belonging to them, such as protein origin, gene name, etc. (Note: The numbers are used to indicate the subclasses of the protein family and the number of proteins, e.g. (12/187): 12 is the number of subclasses of the protein-binding activity modulator family, 187 means that a total of 187 proteins belonging to this family have been collected.)
Click on the protein name on the Upper left corner to be redirected to this protein detail page, which lists the protein, phosphosite and site kinase, functional score and other information in detail.
The functional phosphosites refer to the phosphosites with regulatory information from PSP database, including regulating molecular functions, biological processes, and molecular interactions. Here we present FuncPhos-SEQ, a web server for functional assignment of proteomics phosphosites with a user-friendly interface. FuncPhos-SEQ could predict no biased general functional score, and specific functional (activity or molecular association) score, respectively. In FuncPhos-SEQ, the protein sequence features based on CNN network extraction, evolutionary features of phosphosites and protein network embedding features extracted using SDNE were integrated in heterogeneous feature combination sub-net to make the final functional prioritization.
The FuncPhos-SEQ requires a modern web browser with JavaScript and cookies enabled. To view the complex details, pop-ups must not be blocked. The following browsers have passed the FuncPhos-SEQ test:
This tutorial shows the user how to use the FuncPhos-SEQ to predict phosphosite function.
Click on Tools, FuncPhos-SEQ in order to enter the function prediction page.
The website can be used for phosphosite general function as well as specific function prediction (e.g. Activity and Molecular association), distinguishing between residue types. Users can click on the drop down box to select as required. If activity (ST) is selected, it is used to predict the probability that a phosphosite with residue type S or T in the protein sequence would have activity function.
Two input methods are provided: text box input or loading a sequence fasta file (please use the standard format).
Based on the protein information, the user needs to manually enter one or more phosphosites that they want to predict. The format of the input phosphosites should include residue type and sequence number, such as S114, T262, S293, S391, S384 in the example. Be careful only S/T or Y residue type is permitted depending on the selected prediction models.
Click on the Start prediction button to predict the function of the phosphosites for the selected residue type on this protein sequence.
This is some placeholder content for the scrollspy page. Note that as you scroll down the page, the appropriate navigation link is highlighted. It's repeated throughout the component example. We keep adding some more example copy here to emphasize the scrolling and highlighting.
Threshold value: The user can manually adjust the threshold size. If the site predicted functional score is greater than the threshold value, the function is considered to be present and is displayed in orange on the protein sequence.
The forecast results are also presented in tabular form.
Click the “Save the prediction result to a file” button to download the predicted functional scores of the phosphosites.
Note: This study collects proteins with functional phosphosites from PSP database and calculates their features, rather than all human proteins; Based on a user-entered protein name, the database is looked up and if no features are calculated for that protein, the phosphosite function prediction is performed using the One-Hot channel only.
Phosphorylation modifications play important regulatory roles in many biological processes, and studying their functions is key to understand their regulatory mechanisms. Here we provide a new prediction and visualization platform, FuncPhos-STR. This web application calculates the sequence, structure, and dynamic features of phosphosites based on the protein AlphaFold structures, which enables the prediction of phosphosite function. It also provides users with an intuitive interface to visualize the results, including protein structure, functional scores, and protein features.
The FuncPhos-STR requires a modern web browser with JavaScript enabled. To view the complex details, pop-ups must not be blocked. The following browsers have passed the FuncPhos-STR test:
This tutorial shows the user how to use the FuncPhos-STR to predict phosphosite function.
Click on Tools, FuncPhos-STR in order to enter the function prediction page.
The website can be used for phosphosite general function as well as specific function prediction (e.g. Activity and Molecular association), distinguishing between residue types. Users can click on the drop down box to select as required. If Activity (ST) is selected, it is used to predict the probability that a phosphosite with residue type S or T in the protein would have activity function.
The user enters the protein UniProt ID here. Based on the protein UniProt ID, the site will automatically load the corresponding protein AlphaFold structure.
Based on the protein information, the user needs to manually enter one or more phosphosites that they want to predict. The format of the input phosphosites should include residue type and sequence number, such as S114, T262, S293, S391, S384 in the example. Be careful only S/T or Y residue type is permitted depending on the selected prediction models. In addition, considering the reliability of the predicted protein structure, the user needs to enter the phosphosites with pLDDT greater than 50. For phosphosites with pLDDT less than 50, FunsPhos-SEQ server is recommended.
This is where the user clicks the button to start loading the sample.
Click on the Start to prediction button to predict the function of the phosphosites on this protein.
When feature extraction as well as model prediction is complete, the site jumps to the results page.
On the left side is the AlphaFold structure of the protein, users can adjust the style and apply it. The top table on the right side shows the experimental information, including protein, prediction model, running time, etc. Meanwhile, users can download the result file through the "Download Report" button. The bottom right table shows the predicted results, and the buttons in the rightmost column of each row can control the visualization of the current site in the protein structure picture.
Threshold value: The user can manually adjust the threshold size. If the site predicted functional score is greater than the threshold value, the function is considered to be present and is displayed in red on the protein sequence.
The structural and dynamics features used in StrNet module of FuncPhos-STR, for the prediction of function, activity and molecular association for input phosphosites.
| Function | Activity | Molecular Association |
|---|---|---|
| ANM >60_per_cc | ANM >60_per_cc | ANM >60_per_cc |
| GNM_top20_eigenvectors | Unweighted Closeness | GNM_top20_eigenvectors |
| Unweighted Closeness | GNM_top20_eigenvectors | Unweighted Degree |
| Unweighted Degree | Node-weighted degree | Node-weighted Closeness |
| Page_rank_Iij | ANM_stiffness | Page_rank_Iij |
| ACC | nAA_24_180_pae_smooth10 | ACC |
The annotations of structural and dynamics features.
| ANM >60_per_cc | The average value for each residue with all residues in the cross-correlation matrix based on > 60% ANM modes. |
|---|---|
| GNM_top20_eigenvectors | The eigenvector centrality value for each residue in top 20 GNM modes. |
| Unweighted Degree | Unweighted degree of each residue in its node-weighted amino acid contact energy network (NACEN). |
| Page_rank_Iij | The page rank of each node in PScN. |
| ACC | Solvent accessible surface area for a given residue based on all atoms in the residue. |
| Unweighted Closeness | Unweighted closseness of each residue in its NACEN. |
| Node-weighted degree | The node-weighted degree of each residue in its NACEN. |
| ANM_Stiffness | The average value of stiffness matrix calculated using ANM modes. |
| nAA_24_180_pae | The radius is 24 Å and angle is 180˚ in prediction-aware part-sphere exposure (pPSE). |
| Node-weighted Closeness | The node-weighted closseness of each residue in its NACEN. |
The line chart of the structural and dynamics features for residues in protein structure, with input phosphosites labeled as red dot and with residue type and number information.
School of Biology & Basic Medical Sciences, Soochow University Address: 199 Ren-AiRoad, Suzhou Industrial Park, Suzhou, China PostCode: 215123 Email: zjliang@suda.edu.cn, zhufei@suda.edu.cn