We developed and implemented a method for integrating structure and expression data from many instances of a biological system in order to generate a reliable quantitative description.
First, generate a structure model using our software to obtain a parametric description of the structure or morphology of the organ system or body system of interest. Repeat this operation for different experiments, i.e. different instances of the same biological system. Typically, 5-10 organs or organsims are required for generating later a reliable mean model. The steps to perform are explained in detail in the user manual of our software. For quick starts, they have been summarized in this flowchart.
Once a structure model has been identified using either the unsupervised or supervised methods we developed, save it to file along with a (mean) projection, which can be seen as capturing a snapshot of the gene expression in the entire organ or organism at a given time. With the help of the community, the same approach can be extended to the third spatial dimension and even the fourth dimension (it would no longer be a snapshot of gene expression but a video).
After exporting the structure model and the expression projection for each individual system, click on Expression and select Mean models.
Start by setting the Experiments root directory which contains the directories of every experiment. To easily select the structure model and expression projection from multiple experiment directories, the user can use regular expression (regex).
The following screenshots aim to provide clear examples of how regex can be be used to target different files.
Click on the image to enlarge it. Use the prev/next arrows to navigate between the examples. Examples are discussed in their respective caption. Finally, please refer to the user manual for more information on the required file organization.