The classes and object constructors defined in circlizePlus are prefixed with “cc,” to avoid naming conflicts with other packages and represent the abbreviation of circlize. circlizePlus defines 12 R S4 classes, 7 of which are primary classes, namely, ccPlot, ccPar, ccTrack, ccTrackGeom, ccLink, ccCell, and ccCellGeom. Based on these primary classes, circlizePlus constructs a set of addition operation rules as follows. c c P l o t c o n t a i n   n   c c P a r s + c c P a r = c c P l o t c o n t a i n   n + 1   c c P a r s , n ≥ 0 c c P l o t c o n t a i n   n   c c T r a c k s + c c T r a c k = c c P l o t c o n t a i n   n + 1   c c T r a c k s , n ≥ 0 c c P l o t c o n t a i n   n   c c L i n k s + c c L i n k = c c P l o t c o n t a i n   n + 1   c c L i n k s , n ≥ 0 c c T r a k c o n t a i n   n   c c T r a k G e o m s + c c T r a c k G e o m = c c T r a c k c o n t a i n   n + 1   c c T r a c k G e o m s , n ≥ 0 c c T r a c k c o n t a i n   n   c c C e l l s + c c C e l l = c c T r a c k c o n t a i n   n + 1   c c C e l l s , n ≥ 0 c c C e l l c o n t a i n   n   c c C e l l G e o m s + c c C e l l G e o m = c c C e l l c o n t a i n   n + 1   c c C e l l G e o m s , n ≥ 0

In ggplot2, the parameter data of the function plotting the geometry can be missing. In this case, the value of the parameter with the same name in the function ggplot () will be taken as the default value. Similar features are also implemented in circlizePlus. The coordinate parameters (such as x, y) of the function that draws geometric figures in the sector can be missing. circlizePlus will extract the data of the corresponding sector of the track it is added to and set it as the default value. It is worth noting that the coordinate parameter can be an anonymous function of the form “function (x, y) …”. This anonymous function will be called with the default parameter values above, and its re-turn value will be used as the actual value of the coordinate parameter.

ccPlot is the core class of circlizePlus, whose objects are generated by the homonymous object constructor ccPlot (). The object of ccPlot acts as a container, holding the objects of ccPar, ccTrack, and ccLink. The objects of ccPar are generated by the homonymous object constructor ccPar(), defining the global parameter for plotting. Users may add one or more ccPar objects to a ccPlot object. circlizePlus aggregates multiple ccPar objects, thus determining the global plotting parameters. A single ccTrack object defines how to plot a track. There is also a subclass of ccTrack named ccGenomicTrack. Their object constructors and functionalities are listed in Supplementary Table S1. A single ccLink object stores data of the link connecting two sectors. There are also 2 subclasses of ccLink: ccHeatmapLink and ccGenomicLink. Their object constructors and functionalities are listed in Supplementary Table S2.

When plotting, users are supposed to call the generic show () function, in which ccPlot objects serve as a parameter. Once the show () function is called, circlizePlus will set global plotting parameters based on ccPar objects stored in ccPlot. Meanwhile, it plots tracks based on stored ccTrack objects or their subclasses, and plots lines based on ccLink objects or their subclasses.

ccTrackGeom and ccCellGeom stored in ccTrack determine the geometrics plotted in the current track. Data stored in ccTrackGeom works on the entire track, while data stored in ccCellGeom works only on a single cell within the track. ccCellGeom objects cannot be add directly into ccTrack. It must first be added to a ccCell object before adding the ccCell object to ccTrack thereafter. ccCellGeom has a subclass named ccGenomicCellGeom. The object constructors and functionalities of ccTrackGeom and ccCellGeom are listed in Supplementary Table S3.

There are approximately 6 steps for users using circlizePlus to plot (Figure 2), with no strict order restriction for steps 1 to 4.

Step 1: When initiating plot programming, users must first call the ccPlot () function to generate a ccPlot object. Users also need to set a mandatory parameter named “initMode,” which can take one of the following 4 values: “initialize,” “heatmap.initialize,” “initializeWithIdeogram” or “genomicInitialize.” Different values of initMode correspond to different plotting scenes. The backend of circlizePlus will call the corresponding initialization functions from the circlize package based on the value of initMode. Therefore, ccPlot () function will have different applicable parameters.

This section demonstrates the workflow of circlizePlus based on a small case (Gu et al., 2014). The code lines are segmented according to the workflow defined in the article (codes are shown in Supplementary File S1). It plots the chromosome bands in the outer circle and the corresponding scatter points in the inner circle (Figure 3). Magnified views of chromosomes 7 and 8 are shown on the left parts of the circles. The entire code process is like solving an addition problem, and the data flow of drawing can be clearly understood. Note that the data mapping technique is used in step 3. The coordinate parameter of the function ccGenomicPoints () is missing. It will take the coordinate parameter value of the corresponding sector from the variable track1. This technique ensures consistency of coordinate data while reducing code duplication.

circlizePlus is implemented using object-oriented programming. It is class- and object-centric, with an emphasis on data encapsulation and behavioral reuse. The resulting figure is ultimately determined by the property values of all objects used for drawing, making it easy to maintain and expand.

circlize, on the other hand, is implemented using traditional process-oriented programming, which has functions and processes at its core, and figures are the result of sequential execution of functions. It's suitable for running step by step, with each step calling a function and the corresponding changes will be made immediately on the figure. However, when the parameters that need to be adjusted are used by functions that have been called in some previous steps, the user often has to rerun all the previous code. For example, Figure 3 can be plotted using either circlize or circirclizePlus. When it comes to adjusting the width of the outermost track, circlizePlus needs to run less code, because circlize needs to rerun the code from scratch. In summary, circlizePlus offers advantages in terms of maintenance and scalability, while circlize is better suited for scenarios that require real-time feedback and single-step commissioning (codes are shown in Supplementary File S2).

In this section, we try to draw pictures from real publications. The literature (Alves et al., 2013) uses a plot similar to Figure 4 to present the VCaP cancer cell line. The outermost ring corresponds to human chromosome ideograms (Figure 4A). It serves as an X-axis to indicate the position of the data of other rings on the chromosome. The inner ring in Figure 4B represents the number of gene copies, and the value of each dot is from the Affymetrix SNP arrays. Its Y-axis range is −1 to 1, where dots from 0.15 to 1 are marked in red, dots from −0.15 to 0.15 are marked in grey, and dots from −1 to 0.15 are marked in green. The inner ring in Figure 4C represents the B-allele frequency (ratio), which ranges from 0 to 1 on the Y-axis. The links plot is used to reflect structural variations (SVs) (Figure 4D). The intra- and inter-chromosomal SVs are on the inner rings and depicted in black and red lines, respectively. The red link lines are obscured by the black link lines because there are quite a few intra-chromosomal SVs. The red link lines have been reduced in scale (Figure 4E) for aesthetic reasons. Combining all of the above plots together gives the picture in the literature (Figure 4F).

The process of drawing Figure 4 is to draw the graphs of the various categories first, and then put them together. Quite a few researchers use this process to create circos diagrams. Both circlize and circlizePlus implement such a process (codes are shown in Supplementary File S3). According to code statistics, circlizePlus uses less code than circlize in such a process. Code statistics show that circlizePlus uses less code than circlize in such a process (Table 1).