Semi-supervised identification of SARS-CoV-2 molecular targets

Abstract

SARS-CoV-2 genomic sequencing efforts have scaled dramatically to address the current global pandemic and aid public health. In this work, we analyzed a corpus of 66,000 SARS-CoV-2 genome sequences. We developed a novel semi-supervised pipeline for automated gene, protein, and functional domain annotation of SARS-CoV-2 genomes that differentiates itself by not relying on use of a single reference genome and by overcoming atypical genome traits. Using this method, we identified the comprehensive set of known proteins with 98.5% set membership accuracy and 99.1% accuracy in length prediction compared to proteome references including Replicase polyprotein 1ab (with its transcriptional slippage site). Compared to other published tools such as Prokka (base) and VAPiD, we yielded an 6.4- and 1.8-fold increase in protein annotations. Our method generated 13,000,000 molecular target sequences— some conserved across time and geography while others represent emerging variants. We observed 3,362 non-redundant sequences per protein on average within this corpus and describe key D614G and N501Y variants spatiotemporally. For spike glycoprotein domains, we achieved greater than 97.9% sequence identity to references and characterized Receptor Binding Domain variants. Here, we comprehensively present the molecular targets to refine biomedical interventions for SARS-CoV-2 with a scalable high-accuracy method to analyze newly sequenced infections.

Predicting Epitope Candidates for SARS-CoV-2

“…Additionally, even though a complete analysis with the latest variants of concern was out of the scope for this paper, in light of the outbreak and seriousness of Delta variant of SARS-CoV-2, we got 2 frequently seen sequences of Spike glycoprotein for Delta variant and checked if our top Spike T-Cell and Spike B-Cell epitopes are present on this Spike sequence. The sequences were obtained from 36 and are included in Supplemental files SD1. As seen here 36 , these sequences have been found in other variants of concern as well.…”

“…For this study, we utilized ∼28K unique SARS-CoV-2 protein sequences and ∼49K related protein domains and annotations identified from a collection of ∼62K high quality SARS-CoV-2 genomes from NCBI GenBank 39 and GISAID 40 using the methods described in Beck, et al 36 (full protein and domain sequences can be found with that publication). The proteins and domains identified by FGP are related to the respective source's genome accession.…”

“…The Functional Genomics Platform (https://ibm.biz/functional-genomics) is a comprehensive database and analytics platform that provides (at the time of this work) ~300,000 pre-annotated bacterial and viral genomes, with over 75 million unique gene sequences, over 57 million protein sequences and over 263 million functional domains. For SARS-CoV-2, it contains high quality, curated genomic data defined as genomic sequences with <1% N’s (unknown bases), over 29000 bp and <0.05% unique amino acid mutations (as described in 36 ) along with gene, protein, and functional domain sequences as well as KEGG, IPR and GO codes for those genomes. Moreover, it also provides analytical capabilities such as BLAST over the data.…”

“…Coronaviridae epitope data was retrieved from the Immune Epitope Database and Analysis Resource (IEDB) on July 20th 2020 34 . The IBM Research Functional Genomics Platform (FGP) 35 with updated SARS-CoV-2 genome annotation method 36 was used to identify and retrieve the protein sequences, domain sequences and genome accessions from January 2020 to August 2020. Even though the data from early on in the pandemic, we show the relevance of results, specifically the stability of top 10 epitopes, for newer variant of concern, Delta, as well.…”

“…All epitope data can be obtained directly from iedb.org. All protein sequence data can be obtained from either of the following sources: Functional genomics platform (ibm.biz/functional-genomics). Publication regarding FGP’s annotation of SARS-CoV-2 data 36 . …”