|
|
||
|
|
||
|
- Author: Matteo Ramazzotti - e-mail: matteo.ramazzotti@unifi.it - download: XYLab (Win32 exe) XYLab (linux exe) - source code: code - manual: Manual (.pdf) - program snapshot: the XYLab - test dataset: Escherichia.txt - --- Please visit the XYLab2 page here--- |
||
|
|
||
Maybe the best definition for XYLab is "An interactive XY plotter for multivariate data" and it is dedicated to all the scientists that
work with high throughput techniques (e.g. microarray, proteomics, bioinformatics) that emit their results with overcharged dimensions. It is well known that the so called XY scatter plots are perfect for identifying relationships among variables, but usually it is considered a tedious task to plot everything against everything just to see if something correlates. Not with the XYLab, and this is the reason for which it is born. In XYLab all the plottable variables (i.e. columns containing numerical values) are available as a list in the axes selector: just change the columns and the plot will be readily updated. Clearly the XYLab does much more than this. It has another column selector for the display of a third variable. No, the XYLab has no 3D plotting capability. When the mouse is placed over a plotted point, a label pops-up showing the value of the corresponding entry in the third variable. The management of this third variable is the true innovation of the XYLab. You have the control of a value type that is not bound to numbers, but can contain any type of information, i.e. both numbers and free text. And what can you do with this third variable? For example you can search inside it with a verbose query. Let me explain this with an example: You have plotted a gene expression value (Y) against a predicted index of gene expression (X). And you have selected the "Annotation" variable as label column. With the XYLab you can search inside annotations and evidence all "ribosomal" genes. The results will be highlighted by changing the points appeareance, e.g. they will grow in size and change in color, from black to red. So you have identified in the plot the genes you're interested in. Isn'it a pretty desirable function?. We called it "search-in-plot". Now, let me tell you more about the XYLab. You are certainly used to zoom in/out the plot to evidence some interesting regions. You usually do this by designing a rectangle with your mouse around the region you're interested in. You can do this with the XYLab, too. But with the XYLab you can use the third variable (remember?) in a very useful way: switch the program from "zoom mode" to "copy mode" and the points you selected with your zooming rectangle will be displayed in the copy area of the program in terms of the X, Y and Label coordinates. Nice, isn'it? We called this "select-and-paste". Many other features have been introduced in the XYLab, and they are all detailed in the manual shipped with the program. As you perhaps realized, the XYLab is developed by scientists for scientists. It is not so graphically appealing, right, but it demonstrated in our hands to be sufficiently powerful and definitely useful. Try our XYLab, and let us know your comments, suggestions, and the bug you'll certainly find. It has never been released previously. So be patient and let us improve it to our best. In conclusion, what is XYLab? First of all, the XYLab is not intended to be compared with statistical softwares. It means that it cannot be compared with packages like R, SYSTAT, SPSS, SAS. They really are different things. XYLab is a visualizer, no further ambition. They have superior graphic capability, but they require experience and a scripting skill that is not common among experimentalists. On the contrary, the XYLab is very simple, with an intuitive graphical interface: in brief, it has been written to be simple and to accomplish a single task, i.e. to integrate text based variables in a bi-dimensional scatter plot. Nevertheless, to the best of our knowledge, no other plotting utility allows to identify, once plotted, some desired points among thousand of others, at least not with the same precision. Graphically speaking we stated that the XYLab can monitor three variables at the same time. Other programs can monitor up to five variables in a single plot. For example, with Graphis from Kylebank Software, it is possible to create a 3D scatter plot with size and color of points associated to two other variables. With Microcal Origin one also has the possibility of rotating the 3D plot, therefore increasing its readability. Well, this is not what we wanted to do with the XYLab. The question is not "how many" variables, but "what" variables can be plotted. XYLab integrates text-based information in the plot, and is able to "plot" it in some way.  |
|
  Columns selectors allow to change axes and labels on the fly  Another XY plot from the same multivatiate dataset  With a verbose query it is possible to search inside the third variable ("search-in-plot") and highlight the results in the plot  An XY plot with some points highlighted by the "search-in-plot" function  This switches the mouse behaviour from zoom to copy. In copy mode you use the "select-and-paste" option.   This is the area of the XYLab where the points can be analysed in terms of variable coordinates.  When in copy mode, the coordinates of each point you select is pasted in the paste area.  The XYLab menu. There are several save options, among which the possibility to recreate a subset of the full dataset with currently displayed points. |