The data were preprocessed in the same manner as in [6] except for the last step—the scaling to a zero mean and unit standard deviation. Scaling each variable separately as in [6], however, removes the absolute differences in expression between individual genes. Moreover, it gives the same importance to all genes, including those whose expression is near noise levels, from the point of view of the error metrics. To keep the information about differences in expression levels, we scaled the data by transforming the whole data matrix to have zero mean and unit standard deviation without taking into account that there are different genes—thus, the more expressed genes will be proportionately more expressed even after the scaling. We believe that such scaling is more suitable in this case as the minimization of the error metrics during the fitting phase gives relatively higher importance to more expressed genes and less to the genes whose expression is near the noise level.

Note: The content above has been extracted from a research article, so it may not display correctly.

Please log in to submit your questions online.
Your question will be posted on the Bio-101 website. We will send your questions to the authors of this protocol and Bio-protocol community members who are experienced with this method. you will be informed using the email address associated with your Bio-protocol account.

We use cookies on this site to enhance your user experience. By using our website, you are agreeing to allow the storage of cookies on your computer.