|ModPred: predictor of post-translational modification sites in proteins|
|Home | Help | Datasets|
Frequently Asked Questions
What is the MATLAB Runtime Compiler (MCR)? Top
For users, that do not have MATLAB installed on their computers, the MCR is a shared set of libraries that allows for the execution of programs written in MATLAB. It should be installed and accessible to ModPred and ModPredUI in order for them to run.
I have MATLAB installed on my computer. Do I need to download and install the MCR? Top
It depends on which version of MATLAB (more specifically, which version of the MCR) you have installed on your computer. If the version is the same one as the one in which ModPred was compiled, then you do not need to download and install the MCR. Otherwise, you will need the MCR version that we've provided for ModPred and ModPredUI to run.
When I start ModPred or ModPredUI, I get the following error message: 'Cound not find version 7.XX of the MCR. Attempting to load mclmcrrt7_XX,dll.'. Why does this happen? Top
This error typically occurs when ModPred or ModPredUI cannot find the MCR. This can be due to:
Why does nothing happen for a long time, when I first run ModPred or ModPredUI? Top
The first time ModPred is run, it may take up to a few minutes for the program to execute, without anything happening on screen. This is because the MCR takes some time to extract relevant files and initialize during the first execution. On a 64-bit Windows machine, with a Core2 Duo processor (3 GHz) and a RAM of 4 GB, startup during the first run takes about 90 seconds. During subsequent runs, this reduces to about 20 seconds. These times will vary depending on the make, usage and age of the computer being used.
What does the error message 'Cannot open the ModPred output file. Please try again. If problem persists, contact the authors.' mean? Top
Every time a user submits a sequence to the prediction server it runs the ModPred executable, processes its output and presents it to the user. This generates intermediate files. In some unforeseen situations (like a change in server settings or lack of disk space), these intermediate files cannot be written and the prediction program cannot proceed further. These situations are when the above errors typically occurs. The best thing to do is to try again later.
How current is the information on known PTM sites displayed in the results page? Top
When a site is reported to be known, it is extracted from the training datasets that we have used for ModPred. We do not update this data and it will not remain current.
For how long are prediction results maintained on the server? Top
Prediction results (.txt files) are maintained on the server for a period of 48 hours, during which they can be accessed by simply accessing the link to the results file. After this period, they will be deleted.
Why does each PTM type have different cutoffs for the definitions of low-confidence, medium-confidence and high-confidence sites? Top
ModPred consists of 34 predictive models for 23 types of PTM sites and each of these are trained separately. As a result, each of these models have different performances and thus, different sensitivities and specificities. Therefore, each model has its own set of cutoffs.
If sites do not fall within any confidence definitions, does it mean that they are unmodified? Top
No, ModPred's definitions of confidence are based on standard performance measures used in machine learning and mainly serve as a guide to the user. However, this does not exclude the possibility of a site being assigned a score lower than the pre-defined thresholds but higher than the mid-value (0.5). For example, a disulfide linkage site may be assigned a score of 0.6 which is lower than the low-confidence threshold for this PTM. However, its score is greater than 0.5 and this justifies the possibility of it being an actual site. As a general recommendation, it is worth considering the actual scores provided and defining your own cutoffs for further analyses. However, note that the more relaxed your cutoffs, the more likely you are to include false positives.
Does ModPred predict PTM sites, irrespective of the type of input protein? Top
Yes, ModPred makes NO assumptions about the protein sequence being input. For example, it will predict PUPylation sites on eukaryotic sequences, even though this PTM is almost exclusively present in prokaryotes; it will also predict sulfated tyrosine residues in cytoplasmic proteins, even though it is known that this modification occurs only on proteins that pass through the golgi apparatus. It is expected that, when using ModPred, users have some idea of what PTMs can and cannot occur on their proteins of interest.