MethMarker Tutorial - HTML Tutorial 3 - Step 3/3

Classify Samples

MethMarker opens the biomarker summary window in which you can see the biomarkers found in the PMML models. In our case, it is just one biomarker candidate PY_42.

The regression formula of this biomarker candidate can be seen on the tab Biomarker Performance Summary.

NOTE: Since there are no sample information in the PMML file, there is no training statistic and test statistic at this stage.

Now, go to the tab Classification of New Samples and paste the samples you want to classify in the table in the middle of the window.

CpG methylation grades should be obtained by pyrosequencing, since PY_43.pmml is a pyrosequencing biomarker candidate (also possible would be CO (COBRA), SN (SNuPE), MSP (MSP), ML (MethyLight) or MD (MeDIP)).

After entering the data, click on "Classify".

TIP: Copy and paste the sample information from Excel.

NOTE: The column "class" is not important for this stage. You can leave it as it is, so that MethMarker will assign the class unknown to the samples. Anyway, you can assign individual classes to the samples (methylated or unmethylated). Then, MethMarker will not change the samples' classes.

MethMarker will now classify the samples according to the logistic regression model and report the results.

A classification plot shows the predicted classes of the samples.

A classification table reports the samples, the original classes of the samples (as entered before), the biomarker scores, the predicted classes and the probability for the predicted classes.

Save the classification result as PDF by click on "Save as...".

<< back | next >>