**S**ecure **M**ulti-**P**arty **C**omputation (SMPC) allows multiple parties to compute a publicly known function while keeping their inputs private. This privacy property is desirable, e.g., in the context of distributed **G**enome-**W**ide **A**ssociation **S**tudies (GWAS). In GWAS, researchers try to find associations between variations in genomes and traits (e.g. diseases) by computing certain statistics on sequenced genome data of case and control groups. Since genome sequencing is still comparatively expensive and meaningful GWAS results require large participant groups, it is reasonable for research institutes to collaborate and share their data sets for this purpose. However, participants do not want their genome data to be given out of hand, since data leakage could lead, for example, to discrimination by health insurance companies.

ABY is a widely used framework for compiling function descriptions into highly efficient privacy-preserving protocols in the SMPC special case of only two parties. Therefore, it could be employed to create protocols for securely computing the statistics needed for GWAS conducted by two collaborating research institutes. Using the framework, functions can be described as circuits by plugging together different kind of high-level gates (e.g. ADD and MUL) that operate on encrypted or secret shared input data. There exist arithmetic gates for integer and IEEE 754 floating point input values. However, currently only one type of gate can be used throughout a computation, i.e. for integer inputs only integer operations and likewise for floating point inputs only floating point operations can be performed. This is not desirable in cases where floating point operations on integer inputs are only necessary in the final parts of a computation since the complexity of floating point gates leads to much higher run-times.