Adding 3D structure support to immuneML

Abstract

Adaptive immune receptor repertoire (AIRR) data is used for research into the immune system. immuneML is a machine learning platform used for analysis of adaptive immune receptors. Previously immuneML only supported data in the AIRR1 format. The AIRR format presents the immune receptors as a flat sequence. In reality the immune receptor binding takes place in the 3D space. Running analyses on 3D structure files might prove beneficial. This project expands immuneML’s capabilities by adding PDB2 file support. PDB files can now be used to create datasets in immuneML. This enables users to utilize the information that 3D structures provide, such as the distance and the position of amino acids. Numerous classes have been made for handling the user provided PDB files and creating a dataset object that is suited for storing the information inside. There has also been implemented a few functionalities to show the potential of work that can be done with the data from the PDB files. This project lays the foundation for future analysis of 3D AIRR-data with immuneML when more suitable data is available, which may assist future implementations of 3D AIRR machine learning methods.