The family Caliciviridae consists of a genetically diverse group of single-stranded RNA viruses that can be divided into 10 genera [1]. Of these, viruses belonging to the genus Norovirus and Sapovirus cause acute gastroenteritis (AGE) in humans. Noroviruses are associated with an estimated 70,000–200,000 deaths annually [2,3] while sapoviruses primarily cause sporadic AGE in young children although outbreaks in all age groups have been reported [4,5].

Noroviruses are genetically divided into 10 genogroups and 48 genotypes [6] with viruses in GI, GII, GVIII and GIX infecting humans. Sapoviruses can be classified into up to 19 genogroups (GI–GXIX) [7,8] of which viruses from GI, GII, GIV and GV infect humans [4]. Recombination events within the norovirus genome are well-documented forces that drive norovirus evolution [9,10] and most frequently occur at the junction of the RdRp (polymerase) and VP1 (capsid) encoding regions [11]. Therefore, dual typing of partial regions of both genes is increasingly used for genotyping of norovirus strains [6,12]. For routine typing of norovirus, short nucleotide regions at the 5’-end of the capsid gene and at the 3’-end of the polymerase gene and for sapovirus the 5’-end of the capsid gene are used [13–15]. Traditionally, sequences are compared to an established set of reference sequences to determine genotypes and polymerase types based on phylogenetic clustering, or by employing NCBI BLAST web service. The web-based Norovirus Typing Tool (https://www.rivm.nl/mpf/norovirus/typingtool), which uses a BLAST algorithm against a set of reference sequences followed by phylogenetic analysis to assign norovirus genotypes and polymerase (P)-types and sapovirus genotypes, has been widely used for norovirus and sapovirus typing since 2011 [16].

In this paper, we describe the development and evaluation of an alternative web-based tool, Human Calicivirus Typing (HuCaT) tool for typing of human norovirus and sapovirus sequences. Like the Norovirus Typing Tool, it uses the most recent norovirus and sapovirus reference sequences and nomenclature [4,6,8]. HuCaT uses an algorithm based on matching nucleic acid k-mers (short substrings of DNA [n = 9 nt in HuCaT]) to build alignments between query and reference sequences. With this new approach, we demonstrated the HuCaT algorithm offers speed improvement without losing accuracy for norovirus and sapovirus typing. The HuCaT output includes taxonomic classification (genotype and P-type), percent nucleotide identity, an alignment and phylogenetic tree of query sequence with the closest 12 reference sequences, summary reports, and visualization of query sequences submitted.

To optimize the performance of HuCaT, we calculated the time it took to process 52,176 norovirus and sapovirus sequences downloaded from GenBank. Of these, 6155 norovirus sequences that contained both polymerase and capsid typing regions and 320 sapovirus sequences could be typed by HuCaT. All sequences were queried using different size of k-mers. We settled on a k-mer size of 9 (Fig. 1, Fig S1,b), which resulted in typing of all norovirus and sapovirus viruses in 17 s. In contrast, the BLAST-ref algorithm took 150 s to type all sequences.

Validation of HuCaT using 1299 unique norovirus and 239 sapovirus sequences (Table S1a,b) showed that, except for one GI.P3 sequence (2019-SP-0093), HuCaT accurately assigned all sequences into C-types (genotypes), GII.4 variants and P-types while the Norovirus Typing Tool correctly typed 99.9 % of norovirus genotypes (except GIV.NA1) and 99.0 % of norovirus P-types (Table 2) except one GI.P3, two GI.P11, four GII.P4, one GII.P7, one GII.P17, one GII.PNA6 and one GIV.PNA1 sequence (Table 3). For typing of sapovirus, the Norovirus Typing Tool typed 10 phylogenetically confirmed sapovirus GII.8 sequences as GII.7 and two GII.NA1 sequences as GII.5 (Table 3).

In addition to typing norovirus and sapovirus sequences, HuCaT provides reports on individual (Fig. S3a) or batch submission (Fig. S3b) of query sequences. Single queries generate a report providing the query sequence name and a visual representation of the typing region location (orange [P region] and green [C region] superimposed rectangles) relative to the query sequence (teal rectangle). Additional information includes strand sense (plus/minus), genus, genotype and/or polymerase type (providing the dual typing nomenclature when available), and the percent nucleotide identity with the top matching reference sequences for each typing region. Finally, visualizations are provided for the optimal arrangements (alignments). The 12 reference sequences with the highest similarity are used to illustrate the phylogenetic relationship with reference sequences using UPGMA trees. A close-up view shows the nucleotide differences between the query and reference pairs. Multiple queries are reported initially in a table and include a visual representation of the location of the typing regions similar to the output for single queries. A link is also provided to submit query sequences directly to NCBI BLAST together with highlighted ambiguous residues. Both HuCaT and the Norovirus Typing Tool use the same set of reference sequences. Future versions of the HuCaT tool will include additional sequences of animal noro- and sapoviruses as well as options for exporting reports into different file formats.

The HuCaT tool is a fast web-based tool for accurately typing human norovirus and sapovirus sequences (https://norovirus.ng.philab.cdc.gov). It uses the same set of reference sequences as the Norovirus Typing Tool (https://www.rivm.nl/mpf/norovirus/typingtool) [16] which has been extensively used globally for norovirus and sapovirus typing but uses a BLAST-ref algorithm with reference genomes for typing. HuCaT is much faster than BLAST-ref or phylogenetic clustering approaches without compromising typing accuracy (Table 4). HuCaT is hosted on servers at CDC and is updated when new reference sequences are identified. HuCaT simplifies norovirus and sapovirus typing and generates detailed reports including alignment scores, visualizations and phylogenetic trees.

HuCaT has several limitations that will be addressed in the next version of the tool. First, only query sequences that include the entire typing region can be typed accurately, and shorter sequences and those with indels are not typed. Also, sequences with >13 % nucleotide difference compared to references sequences cannot be typed. Hence, if there are significant regions with poor sequence quality, the sequence may be untypeable. Future updates to the algorithm will account for sequence quality problems or error messages will describe why a query is untypeable. Including error messages would also remove the ambiguity when a strain is deemed “untypeable”, reserving the term for unique strains rather those that more likely contain sequencing errors. Inherent limitations of any sequence-based typing tool that uses relatively short sequences for typing, include the possibility of mistyping a novel recombinant virus or a unique strain that has differences outside the typing region.

Robust typing tools are needed for standardized genotyping of highly diverse RNA viruses such as noroviruses and sapoviruses. There are a tremendous number of bioinformatic tools that use k-mer vocabulary, including alignment-free classification tools that use k-mer frequency and distance calculations to infer phylogeny [14]. Such methods are much faster than alignment-based k-mer methods, such as BLAST, but positional information on the similarity/dissimilarity of query and reference sequences is lost without a sequence alignment. The HuCaT algorithm starts with an alignment free approach, but simultaneously builds an alignment of query-reference pairs for calculating nucleotide percent identity. As a result, HuCaT overcomes the loss of information of other k-mer based tools and accurately types input sequences.

In conclusion, genotyping of norovirus and sapoviruses is important to monitor epidemiological trends of outbreaks and sporadic illnesses. HuCaT adds to the availability of tools for typing the highly divergent human noroviruses and sapoviruses. Together with the Norovirus Typing Tool, HuCaT includes the latest updates in new norovirus genogroups, genotypes and P-types [6].