ddalg

Algorithms and data structures for my Python projects.

Interval tree

Interval tree is a data structure for holding intervals and to allow efficiently find intervals that overlap with given interval or point. Read more on Wikipedia.

Implementation note

This implementation uses half-open intervals, where begin coordinate is excluded. Half-open intervals are used in e.g. BED genomic format.

The current implementation needs to rebuild the tree after each insert, hence the tree is not efficient for using in read/write fashion.

Usage

• implement your custom interval object while extending Interval. Two properties need to be overwritten:
  • begin - 0-based (excluded) begin coordinate of the interval
  • end - 0-based (included) end coordinate of the interval

    from ddalg.model import Interval
    
    class YourInterval(Interval):
    
      def __init__(self, begin: int, end: int):
        self._begin = begin
        self._end = end
        # anything else
    
      @property
      def begin(self):
        return self._begin
    
      @property
      def end(self):
        return self._end
    

• create a collection of your intervals and store them in the interval tree:

  from ddalg.itree import IntervalTree
   
  itree = IntervalTree([YourInterval(0, 3), YourInterval(1, 4)])
  # ... 
  

• query itree:
  • using 1-based position:

    itree.search(1)
    

    returns (0,3)

  • using half-open interval coordinates:

    itree.get_overlaps(0, 1) 
    

    returns (0,3), effectively the same query as above

  • for intervals with minimal required coverage

    itree.fuzzy_query(0, 1, coverage=.90)
    

    return intervals with >=.9 overlap with respect to query coordinates

  • for intervals with minimal jaccard index

    itree.jaccard_query(0, 1, min_jaccard=.90)
    

    return intervals having jaccard_index>=.9 with respect to query coordinates