Biomedical Engineering Reference
In-Depth Information
the top-n rows should be passed). However, in this case some of the
parameters for the node are passed from a different node. The Integer
Input node is part of a collection of so-called quickform nodes that allow
injection of certain variables into a workfl ow from the outside. In the
quickform node a description and a default value for the variable is
entered, for example the number of amines/acids that should be passed
through. When used inside a meta-node all variables defi ned by quickform
nodes are exposed in the dialog of the meta-node, that is the user can
confi gure the meta-node from the outside without opening it. Once values
for each variable are available, they are passed into the row sampling
nodes via special variable ports (red balls instead of the usual white
triangles for data tables). In the sampling nodes' dialog the parameter for
the number of rows is subsequently linked to the variable instead of
taking the (default) value that was directly entered in the dialog.
In combination with the commercial KNIME Server, quickform nodes
can also be used outside meta-nodes. Here, variables are exposed in a
web interface allowing the user to easily confi gure a workfl ow from a
web browser and execute the workfl ow on the server, see Figure 6.11.
The results (fi les, numbers, . . .) are also available via the browser. This
concept allows 'power users' to build elaborate workfl ows and expose
only certain parameters to the outside. Once uploaded to the server
(directly from within KNIME), the 'normal' users can access the
workfl ows via the web portal with the browser, provide input fi les and
parameters, execute it on the server, and fi nally retrieve the results. A
workfl ow can even be executed several times and all results are stored.
6.4.3 Classifying cell images
The third example workfl ow solves a common task in bioinformatics: the
automated classifi cation of cell images. One use case is the treatment of
cells with different substances at various concentrations. At some point
cells start to die and the task is to ascertain the lethal concentration
automatically. For this, cell cultures are grown in 96-well plates and are
treated with the substances in different concentrations. On each plate
there are always positive and negative controls, for example the cell
cultures without additional substances and cultures with very high
concentrations of known activity, respectively.
Automated imaging systems take pictures, one for each well. Using the
Image Processing nodes (which again rely on the free Imglib [6]), which
are also available from the community contributions, these images can be
Search WWH ::
Custom Search