Highlights d PepSeq enables fully in vitro, highly multiplexed peptidebased antibody assays d Epitope mapping shows preexisting antibody reactivity to SARS-CoV-2 antigens d Antibodies cross-recognize endemic and pandemic antigens in the Spike S2 subunit d Cross-reactive antibodies raised by SARS-CoV-2 preferentially bind endemic homologs

A high-resolution understanding of the antibody response to SARS-CoV-2 is important for the design of effective diagnostics, vaccines and therapeutics. However, SARS-CoV-2 antibody epitopes remain largely uncharacterized, and it is unknown whether and how the response may cross-react with related viruses. Here, we use a multiplexed peptide assay (‘PepSeq’) to generate an epitope-resolved view of reactivity across all human coronaviruses. PepSeq accurately detects SARS-CoV-2 exposure and resolves epitopes across the Spike and Nucleocapsid proteins. Two of these represent recurrent reactivities to conserved, functionally-important sites in the Spike S2 subunit, regions that we show are also targeted for the endemic coronaviruses in pre-pandemic controls. At one of these sites, we demonstrate that the SARS-CoV-2 response strongly and recurrently cross-reacts with the endemic virus hCoV-OC43. Our analyses reveal new diagnostic and therapeutic targets, including a site at which SARS-CoV-2 may recruit common pre-existing antibodies and with the potential for broadly-neutralizing responses.

PepSeq is an in vitro platform for building and conducting highly multiplexed proteomic assays against customizable targets by using DNA-barcoded peptides. Starting with a pool of DNA oligonucleotides encoding peptides of interest, this protocol outlines a fully in vitro and massively parallel procedure for synthesizing the encoded peptides and covalently linking each to a corresponding cDNA tag. The resulting libraries of peptide/DNA conjugates can be used for highly multiplexed assays that leverage high-throughput sequencing to profile the binding or enzymatic specificities of proteins of interest. Here, we describe the implementation of PepSeq for fast and cost-effective epitope-level analysis of antibody reactivity across hundreds of thousands of peptides from <1 µl of serum or plasma input. This protocol includes the design of the DNA oligonucleotide library, synthesis of DNA-barcoded peptide constructs, binding of constructs to sample, preparation for sequencing and data analysis. Implemented in this way, PepSeq can be used for a number of applications, including fine-scale mapping of antibody epitopes and determining a subject's pathogen exposure history. The protocol is divided into two main sections: (i) design and synthesis of DNA-barcoded peptide libraries and (ii) use of libraries for highly multiplexed serology. Once oligonucleotide templates are in hand, library synthesis takes 1-2 weeks and can provide enough material for hundreds to thousands of assays. Serological assays can be conducted in 96-well plates and generate sequencing data within a further ~4 d. A suite of software tools, including the PepSIRF package, are made available to facilitate the design of PepSeq libraries and analysis of assay data.

No abstract

As an increasing number of leadership-class systems embrace GPU accelerators in the race towards exascale, efficient communication of GPU data is becoming one of the most critical components of high-performance computing. For developers of parallel programming models, implementing support for GPUaware communication using native APIs for GPUs such as CUDA can be a daunting task as it requires considerable effort with little guarantee of performance. In this work, we demonstrate the capability of the Unified Communication X (UCX) framework to compose a GPU-aware communication layer that serves multiple parallel programming models developed out of the Charm++ ecosystem, including MPI and Python: Charm++, Adaptive MPI (AMPI), and Charm4py. We demonstrate the performance impact of our designs with microbenchmarks adapted from the OSU benchmark suite, obtaining improvements in latency of up to 10.2x, 11.7x, and 17.4x in Charm++, AMPI, and Charm4py, respectively. We also observe increases in bandwidth of up to 9.6x in Charm++, 10x in AMPI, and 10.5x in Charm4py. We show the potential impact of our designs on real-world applications by evaluating weak and strong scaling performance of a proxy application that performs the Jacobi iterative method, improving the communication performance by up to 12.4x in Charm++, 12.8x in AMPI, and 19.7x in Charm4py.