This page provides all information on how to use this website.
You can either sign-up or use a guest submission to upload and process your data.
However, it is not neccessary to sign up in order to use this service.
If you want to have an overview about your submissions we suggest to create an account.
If you just want to test the rMKL-LPP tool we recommend to use a guest submission.

The local executables for Linux, Windows and MacOs of rMKL-LPP is available here.

If you have any questions or need support, feel free to send us an email.

Offline Tool

You can find details for the offline kernel computation tool below.


Prediction Disclaimer

All provided data output is for research use only, not for diagnostic or clinical purposes! We do not guarantee for any prediction!

Help Topics

1. Sign Up 2. Log In 3. Home   3.1 Upload Data   3.2 Choose Settings   3.3 Compute Submissions   3.4 Display Results 4. Submissions 5. User Settings 6. Offline Tool


Browser compatibility
OS Version Chrome Firefox Microsoft Edge* Safari
Linux Ubuntu 18.04 72 65 n/a n/a
MacOS 10.14 Mojave 72 65 n/a 12.03
Windows 10 (1809) 72 65 44 n/a
* Fully functional, but does not support desktop notifications.



1. Sign up (optional)

If you want to have an overview of your submissions you can create an user account.
To sign-up, click on the 'Log-In' button on the top-right corner and then click on 'Sign Up'.
Alternatively you can click here for a direct link.


2. Log In (optional)

To log in, just click on the green 'Log-In' button located at the top-right corner.
After successfully filling out the user name and password you will be logged-in and be able to upload your data and see your submissions.
To log out, click on the 'logged-in' button and then select 'Log-Out'
If you do not have an account already but want to create one, read through step 1. Sign Up.


3. Home

On this page you can upload your data. After that you can choose the prefered configuration and receive the result afterwards.
Please do not upload any patient data. In order to compute a kernel matrix, you can use the provided offline tool below.


3.1 Upload Data

For demonstration of the service you can use the provided example files after downloading them from here.


On the upload page, click on the field and select the data you want to upload.
For the first field you may only select a file in .txt format.
For the second field you can select multiple kernel files. Those have to be in .mat format.
If you only have .csv format files, please use the offline tool to convert it to .mat files.

3.2 Choose Settings

Give your submission a recognizable name and select the settings of your choice.
Submissions with a chosen dimension of 2 or 3 can be visualized after computation online.
To get further information about the parameters, please hover over the red info-logos.
If you are logged in you can not see the field with an email-adress. Submissions and computation status can be seen at submissions. Important! If you are using a guest submission, please provide an email adress, if you want to get notified when your submission is ready.

3.3 Compute Submission

At this point your submission is in the queue or is being computed.
Important! If you are using a guest submission, please copy the link to your clipboard.
Only with this link you can access your submission results within the next 30 days when you close the tab and did not provide an email-adress.

If you are logged in, you can close the browser and revisit at a later stage.
There is also the option to receive either desktop- or email-notifications once the computation is over.
To activate them, please read through section 5. User Settings.
After successfully computing the result, you can go to the results section.

3.4 Display Results

Here you can download the computed data as a zip file. You can also view the parameters that have been set for this submission by clicking the 'View Parameter' button.

If you are logged in you can share you submission by clicking on 'Share Submission'. Anyonw with the link from the clipboard below, can access this submission now. If you want to unshare this submission, you can press the 'Unshare Submission' button.
The data can not be accessed by others anymore.
Important! If you are using a guest submission and want to access the results later, bookmark the link from the clipboard below.
If the setting of dimensions was set to 2 or 3, you can interactively explore the results.
When you move your mouse on the diagram, different tools appear on the top-right corner of the diagram.
With these tools you may zoom in, select or hide certain data points.


4. Submissions (optional)

If you are logged in, this tab gives an overview of your submissions within the last 30 days. For each submission the name, date and state is specified.
If the submission has already been computed, you find a button to directly download the results or go to 'view'. For more information about this, please read through section 3.4 Display Results.


5. User Settings (optional)

If you created an account and you are logged in, you can click on the green 'Logged-In' button and then select 'Settings' to get to preferences.
This page lists information about your account, like how many submissions you currently have on the server.
Here you can also activate the email or desktop-notifications. To get email notifications, simply activate the checkbox.
To get desktop notifications click on 'Set Permission' and then grant your browser the allowance to send desktop notification.
If you want to turn off desktop notification, you need to set the permission in your browser settings. This link explains how to set these permissions in Chrome, Firefox and Safari. With the red button at the end you get the option to delete your account and all associated data.






Offline precomputation tool

Using the offline precomputation tool (download exe or jar) you can either compute kernel matrices from CSV files containing numerical data or import, validate and convert precomputed matrices.
System requirements: This tool requires Java version 11.0.2 and your computer should have at least 4GB of RAM.
To run the jar application, please open the terminal and run:
java -jar /path/to/web-rMKL_preprocessing.jar

Help Topics



Import CSV raw data

Raw data can be imported as a CSV containing samples as rows and features as columns (“Import raw data” menu). The first row should contain column headers while the first column may contain sample IDs ("Data including ID tags" option). All sample data must be numerical data and all rows must be of equal size.
You can import multiple CSV files under the condition that all of them contain the same number of samples. Files that do not fulfill these preconditions cannot be imported.



Compute kernel matrices

After the successful import of at least one CSV file, you can select your desired kernel computation method.
Furthermore, you can choose to either apply default parameters based on the number of features in your dataset or select all available parameters yourself.
The "Apply to all matrices" button will apply the computation parameters of the currently selected matrix to all imported matrices.

Kernel choice:
In general, RBF kernels are good for real values (like DNA methylation or gene expression data) and polynomial kernels are good for discrete data (e.g., genetic data). Linear kernels might give reasonable results, if the data does not contain non-linear relationships.

If you hit the compute all button, all input matrices will be computed. You will be forwarded directly to the "Kernel Matrices" tab as soon as the computation is completed.



Convert kernel matrices

Since the rMKL webserver only accepts MAT files, you have the option to import and convert precomputed matrices from CSV files using the "Import precomputed" button. The files must only contain numerical data and must be symmetrical. After successful import, you will be forwarded to the “Kernel Matrices” tab for validation and export.



View and export kernel matrices

In the "Kernel matrices" tab, you can either view matrices that were computed by this tool or precomputed matrices that you imported from a CSV file. Next to the file name of your selected matrix, you will be notified whether the matrix is valid for use with the rMKL web server (only for precomputed kernel matrices).
For exporting the matrices, you will be asked to specify an export directory where all matrices will be stored as MAT files. In addition to the matrices, a sample ID TXT file is generated.
Note that only valid kernel matrices will be exported. This means that precomputed matrices have to be symmetric.

License: This tool is distributed and may be redistributed under the GPL license (Copyright © 2019, Nicolas Kersten). For Mat file export, matfilerw (Copyright © 2015, DiffPlug) is used.