Federated learning is an important distributed machine learning paradigm. This study proposes a privacy-preserving data augmentation model for federated learning of heterogeneous data, which is able to mitigate heterogeneity and augmenting the participant’s local data while protecting data privacy. First, to address the problem of global model bias due to heterogeneous data, this study proposes a distributed generative adversarial network FedEqGAN. The model introduces a multi-source data feature fusion mechanism, which can learn the features of each data source to generate synthetic data. Second, addressing the privacy leakage issue caused by the disclosure of data distribution information, this paper proposes an encryption algorithm for heterogeneous environments FedHE, which utilizes homomorphic encryption to protect local data distributions and aggregates local data information through KL dispersion in order to construct global data distributions. Finally, for the privacy leakage problem caused by uploading model parameters in federation training, this paper proposes a federation model parameter encryption algorithm DPFedMP. This algorithm dynamically injects Gaussian noise into the model parameters according to the difference of data distribution to realize differential privacy protection and update the global model. Experiments show that the method is applicable to heterogeneous data environment, significantly enhancing model performance while ensuring data security.

MOTIVATION: Ribosome profiling has been widely used to study translation in a genome-wide fashion. It requires deep sequencing of ribosome protected mRNA fragments followed by mapping of fragments to the reference genome. For applications such as identification of ribosome pausing sites, it is not enough to map a fragment to a given gene, but the exact position of the ribosome represented by the fragment must be identified for each mRNA fragment. The assignment of the correct ribosome position is complicated by the broad length distribution of the ribosome protected fragments caused by the known sequence bias of micrococcal nuclease (MNase), the most widely used nuclease for digesting mRNAs in bacteria. Available mapping algorithms suffer from either MNase bias or low accuracy in characterizing the ribosome pausing kinetics. RESULTS: In this paper, we introduce a new computational method for mapping the ribosome protected fragments to ribosome locations. We first develop a mathematical model of the interplay between MNase digestion and ribosome protection of the mRNAs. We then use the model to reconstruct the ribosome occupancy profile on a per gene level. We demonstrate that our method has the capability of mitigating the sequence bias introduced by MNase and accurately locating ribosome pausing sites at codon resolution. We believe that our method can be broadly applied to ribosome profiling studies on bacteria where codon resolution is necessary. AVAILABILITY AND IMPLEMENTATION: Source code implementing our approach can be downloaded under GPL3 license at http://bioserv.mps.ohio-state.edu/RiboProP. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.