PhylogicianJS

Node.js module to visualize phylogenetic trees.

Development rational

Phylogician class responsibility distribution

The Phylogician class represents the tree itself.

We can feed the Phylogician class with newick data using .data(). Phylogician then parse the newick using newick-reader npm package.

The newick-reader returns a deeply nested object with the tree structure.

We decided to flatten this scruture and store each tree node as an array of TreeNodes. Phylogician does that with .makeTree() method.

This makes sense, right? A phylogenetic tree is a collection of nodes. Each node should hold all the information to place it on a abstract version of the tree, for example an array.

Thus, an element of the TreeNode class has the following information:

public name: string
public branchLength: number|null
public children: number[]

private id: number|null
private parent: number | null
private root: boolean

private viz: IvizOptions

We do not exposed id, parent and root directly because they are attributes assigned during the construction of the tree by the Phylogician method .makeTree(). That information is needed because we flattened the nested structure of the original ITree object To access these values, there are getters and setters methods in the instances of the TreeNode class. They are designed to protect the developer from accidentally re-assing these values. In particular id that should be only assigned once. parent and root can be reassinged during reroot for example.

The TreeNode is self contained and our philosophy is that each instance of the class should have all information about the node it represents in the phylogenetic tree. For that, TreeNode also have the private property viz with the following interface:

interface Ipos {
    x: number
    y: number
}

interface Idim {
    h: number
    w: number
}

interface IvizOptions {
    nodeSize: number
    nodeColor: string
    pos: Ipos,
    labelDim: Idim,
}

The IvizOptions store the size of the node, its color, position and dimensions of the label. The TreeNode class has the nodeSize and nodeColor set to 3 and black as default. Both pos and labelDim are set to 0 both for x, y in pos and h, m in labelDim.

These attributes are private but can be set and get with the appropriate public methods.

Ok, now that we have the tree on an abstract set that can be publically accessed by the Phylogician attribute .nodes, it is time to pick a representation with .layout(). This is when Phylogician set the labelDim and pos attributes of each node before it draws it.

There is only vertical layout available right now.

and how to represent this data by picking a layout using .layout().

Only vertical layout is available.

So the tree itself is a collection of TreeNodes and a Layout.

This helps us to distribute responsibility between the classes.

For instance, TreeNode has:

TODO:

• [ ] Fix CI https://medium.com/@ali.dev/how-to-setup-chrome-headless-on-gitlab-ci-with-puppeteer-docker-fbb562cbaee1
• [ ] Tooltip builder (its own class)
  • [ ] Laderizer
  • [ ] branch opacity based on uncertainty
  • [ ] collapse
• [ ] Resolve overlapping names.
• [ ] ReadMe