Fragment Searching across the PDB using MMTF with mmtf-fragment-search

A way of querying all fragments in the PDB (~32 million) for a specific property implementing the recently developed mmtf-spark library.

BioJavaStructureToFragments class

Provides functionality for fragmenting the entire PDB based upon an arbitrary window size.

ConsecutiveFragment class

Filters out non-consecutive peptides using Calc.isConnected from BioJava

SimilarityScorer class

Calculates a very simple metric for assessing the similarity of a query fragment to the target fragment. It is calculated as the sum of differences in angles at each position in the window. (Score of 0 is perfect)

ResultsDataset class

Converts the JavaRDD to a Spark Dataset for displaying and added functionality of sorting/fetching columns

FragmentSearch class

Main class that puts everything together, keeps performance metrics, and prints the desired number of sorted (ascending) fragment hits.