MetalPred help

This page will help you to find your way around MetalPred.


MetalPredator is designed to predict iron-sulfur cluster binding proteins from protein sequence(s). Our tool integrates an existing methodology for domain-based predictions (Andreini et al., 2011) and a new approach to search for metal-binding motifs, based on MFSs or fragments thereof. This integration exploits the complementarity between the global properties of protein domains and the local nature of MFSs and adds an important feature to metalloproteins prediction, namely the ability to predict metal-binding sites in proteins with novel folds. A crucial basis of this is the ability to search for smaller fragments associated with MFSs that are not limited to known metal-binding domains. This constitutes an unbiased approach that allows identifying both known metal sites in novel proteins and new metal sites as a combination of individual metal-binding fragments from different MFSs.

MetalPred home page

MetalPred can process the entire proteome of any organism in minutes to a few hours (e.g. for the human proteome), and thus can be applied to any newly sequenced organism, including eukaryotes.

You can submit the sequence(s) by either pasting them directly as the plain text (1) or uploading as a file (2). The only supported inputs to the server are the amino acid sequences formatted in FASTA format, e.g. the FASTA definition line followed by the line(s) of sequence data.
(3) You can also provide an e-mail address which will be used to send a notification when results are ready.
When ready, hit the Submit button (5) to start the prediction job. Hitting the Clear button (4) will discard all data from the input form.
After the successful submission you will be redirected to the page that contains a dynamically generated link that provides an access to a web page with results.
NOTE: If an email address has not been specified, keep the results link available to access your predictions.

Review MetalPredator results

The results Summary page is the main output page which you can access using the generated results link as soon as prediction is finished. This gives an overview of the all the predicted iron-sulfur cluster binding proteins identified within your input sequences. The sequence ID appears on the left, then from left to right, for each predicted protein the columns of the report contain the results for Domain search with pattern filter (6), MFS search (7), Fragment search (8), and Domain search without pattern filter (9) predictions. You can click directly on the prediction results in the table to open a pop up window with details.
Here are ordered according to the number of methods that predict the binding. When a putative site is found the green color of the corresponding cell provides a visual clue and the cell reports all the predicted binding residues. Otherwise an empty cell is highlighted in red. When the prediction is obtained by Domain search without pattern filter, we report only the Fe-S cluster binding domain identified.
The first 10 entries are returned by default, allowing immediate analysis of the top matches. The remaining results can be viewed by clicking on the pagination links found above and below the table (10).
You can filter the results to find exactly the sequence he/she wants to review using the Filter box (11) in the top left corner of the page. Hitting Clear button will discard the filtering and display all results.
A downloadable csv file is created upon the user’s selection of the columns to be included. Click the Generate Report button (12) in the right top corner of the page and select the methods you want to include. You can choose to download all the results returned by selected methods or you can click on Filtered results to download only the "filtered rows".
Additionally, the interface presents the results for each method on separate tabs and you can switch between them by selecting a corresponding tab at the top of the page (13).
In each tab, the predicted iron-sulfur binding sequences are sorted by e-value (14). You can drag the horizontal scroll, that appears when pointing the sequence column (15), to see where the profile matches the sequence (highlighted in green) and the predicted ligands (highlighted as blue-colored bold letters). Each page has the Download all button (16) which you can use to download the list of identifiers of the sequences predicted by the method. Use the Download filtered button to download only the results that meet filtering criteria (when Filer is used).
There can be multiple hits for one sequence because more than one profile can be matched with a sufficient score. The Details link (17) pops up the list of all profiles matching a given sequence with additional information on the match. For the Domain search with pattern filter method an input sequence is predicted to bind an iron-sulfur cluster if the profile of a domain with associated ligands matches the sequence with an e-value lower than 10-5 and ligands are conserved in the sequence.
This description includes the profile identifiers (18), the bit-score, bias and e-value given by the HMMER tool, the start and end positions of the profile alignment to the sequence (19) and the predicted ligands (20) (not reported for Domain search without pattern filter method). The sequence column highlights the location of putative binding domain and of the corresponding metal-binding residues.
MFS and Fragment searches are local searches for metal-binding motifs. For these two methods an input sequence is identified as a potential iron-sulfur cluster binding protein if:
(i) all fragment profiles of a given MFS (21) match the sequence with an e-value lower than 10-3 and the corresponding ligands are conserved in the sequence (MFS search).
(ii) at least one fragment (22) profile of a given MFS matches the sequence with an e-value lower than 10-3 and the corresponding fragment ligands are conserved in the sequence (Fragment search).
The Domain search without pattern filter method predicts the sequence to be an iron-sulfur cluster binding protein if the profile of a domain with no information on ligands available matches the sequence with an e-value lower than 10-20. No information on a putative metal-binding site is provided.