Publications by authors named "V Popic"
Article Synopsis
- Finding relatives in genomic studies is tough when data is spread across multiple organizations with sharing restrictions.
- SF-Relate is a new federated algorithm that uses a unique hashing approach to efficiently and securely identify genetic relatives by grouping individuals into buckets and only comparing those in the same group.
- It ensures privacy through multiparty homomorphic encryption, allowing secure computation of relatedness without any private data being shared, successfully identifying 97% of close relatives in large datasets like the UK Biobank.
Article Synopsis
- Finding relatives is crucial for genomic studies, but data-sharing restrictions complicate this across different entities.
- SF-Relate is a federated algorithm that uses locality-sensitive hashing to efficiently identify genetic relatives while preserving privacy.
- By using multiparty homomorphic encryption, SF-Relate allows data holders to compute relatedness without sharing sensitive information, achieving high detection rates in large datasets like the UK Biobank.
Gene fusions are found as cancer drivers in diverse adult and pediatric cancers. Accurate detection of fusion transcripts is essential in cancer clinical diagnostics, prognostics, and for guiding therapeutic development. Most currently available methods for fusion transcript detection are compatible with Illumina RNA-seq involving highly accurate short read sequences.
View Article and Find Full Text PDFArticle Synopsis
- Long-read RNA-sequencing methods can capture full transcript isoforms but traditionally have low throughput*. -
- The new technique, multiplexed arrays isoform sequencing (MAS-ISO-seq), enhances this by combining cDNAs for more efficient long-read sequencing, boosting throughput by over 15 times*. -
- In experiments with tumor-infiltrating T cells, MAS-ISO-seq led to a significant increase (12- to 32-fold) in the identification of differentially spliced genes*.
Article Synopsis
- Structural variants (SVs) significantly contribute to genetic diversity and disease, highlighting the need for better detection methods in precision medicine.
- Existing methods for detecting SVs are limited because they rely on manual features and rules, which don't scale well to the wide diversity of SVs in genomic data.
- The Cue framework uses deep learning to analyze sequencing data by converting alignments into images and employing a convolutional neural network to accurately predict SV types, achieving superior performance compared to current methods.