Which approaches traversing through a Typescript AST yield what?

I started experimenting with the Typescript compiler, especially using the ts-morph wrapper. Currently I am attempting to traverse through the AST of a source file to deduce certain kinds of nodes and their properties and noticed at least three possibilities to traverse the respective AST which all yield different results:

• SourceFile.getChildSyntaxListOrThrow().getChildren().forEach(...) in combination with recursive node.getChildren().forEach(...) call which seemingly is the only approach to get trivia and comments as well.
• SourceFile.getChildren().forEach(...) sans .getChildSyntaxListOrThrow()
• SourceFile.forEachDescendant(...) without the necessity to to call a visitNode handler for child node
• SourceFile.forEachChild(...) that does not yield keywords

Unfortunately I cannot seem to understand the exact difference to fully figure which approach is best in which case. So the question is "which approaches traversing through a Typescript AST yield what?"