Retrospective search for SARS-CoV-2 in human faecal metagenomes
Simone Rampelli,1,2,* Elena Biagi,1 Silvia Turroni,1 and Marco Candela1,*
1
Unit of Holobiont Microbiome and Microbiome Engineering, Department of Pharmacy and
Biotechnology, University of Bologna, Bologna 40126, Italy
2
Lead Contact
*Correspondence: simone.rampelli@unibo.it (S.R.) and marco.candela@unibo.it (M.C.)
Main text
Shotgun metagenomics of faecal DNA provided an unprecedented opportunity for a non-invasive
characterization of the human faecal virome, as representative of the complex viral metacommunity
that populates the human gut (Nakatsu et al., 2018). We have recently developed a tool called
ViromeScan that allows for rapid extraction of the total human virome from metagenomic
sequences (Rampelli et al., 2016). ViromeScan has been applied to human faecal metagenomic
datasets, highlighting both the high complexity and the relevant degree of variability of the human
gut virome across the globe (Rampelli et al., 2017).
According to very recent publications, SARS-CoV-2 nucleic acid was recovered in the faeces of
infected patients, even after the oropharyngeal swab turned negative (Ianiro et al., 2020; Xiao et
al., 2020; Xu et al., 2020). This indicates the oportunity to monitor viral shedding also in the
gastrointestinal tract, as this might be greater and more lasting than that in the respiratory tract.
Since several protective elements of the SARS-CoV-2 nucleic acid (Deng et al., 2017; Hagemeijer et
al., 2012; Knoops et al., 2008) might favour the persistence of viral double-stranded RNA in faecal
samples, we hypothesised its recovery along with DNA during DNA-targeted extraction protocols
and subsequent sequencing in shotgun metagenomic approaches. Therefore, in the present study,
we attempted to use ViromeScan to retrospectively search for SARS-CoV-2 genomic traces in human
faecal viromes from publicly available metagenomes. First, in order to allow detection of SARS-CoV-
2 sequences, ViromeScan was implemented by integrating its database with the representative
sequence for the SARS-CoV-2 genome. A second database containing 45 genomes for SARS-CoV-2
and 55 genomes for other coronaviruses, including viruses from other hosts, was then built, on
which the reads assigned to the Coronaviridae family by ViromeScan could be specifically mapped,
to allow the definitive assignment of the sequences to the SARS-CoV-2 genome (Figure 1, see
Supplemental Experimental Procedures for more details). This implemented version of our tool was
finally applied on a total of 38 faecal metagenomics samples, from 26 Chinese elderly women (Wang
Electronic copy available at: https://ssrn.com/abstract=3557962
et al., 2019) and 12 European adults (3 Italians and 9 Spaniards) (Rampelli et al., 2015; Mas-Lloret
et al., 2020). According to our findings, well-defined virome structures are evident for each analysed
metagenome. In particular, we retrieved reads for Anelloviridae, Arenaviridae, Bunyaviridae,
Coronaviridae, Flaviviridae, Hepeviridae, Herpesviridae, Orthomyxoviridae, Papillomaviridae and
Poxviridae, confirming the ability of ViromeScan to detect both DNA and RNA viruses. When
focusing on Coronaviridae, we detected traces of SARS-CoV-2 in 6 out of 26 metagenomes all
belonging to the Chinese cohort, while 10 metagenomes, including 1 Chinese and all 9 Spanish
samples, contained reads for other coronaviruses (Figure 2). It should be noted that in order to
reduce the probability of false positives, SARS-CoV-2 was considered actually present when the
samples contained more than 3 hits and more than 30% of the total reads retrieved from the first
step of the pipeline were assigned to SARS-CoV-2 genomes. SARS-CoV-2 was found to be the most
represented Coronaviridae species among the 6 Chinese samples in which it was detected (mean
relative abundance on the total reads assigned to Coronaviridae ± sem: 38±3%). In addition, these
metagenomes showed a higher overall amount of reads assigned to Coronaviridae than other
samples positive for coronaviruses (mean absolute count ± sem: 21±6 and 13±2 for the 6 Chinese
and the 10 other samples, respectively).
Taken together, the data from this preliminary, exploratory investigation provide evidence to
support the presence of genomic traces of SARS-CoV-2 in the faecal virome of 6 out of 26
presumably asymptomatic Chinese, whose biological samples were collected before April 2019. By
contrast, no SARS-CoV-2 was detected in the faecal viromes from 12 EU citizens, for whom faecal
sampling dates back to March-April 2013 for Italians and before November 2019 for Spaniards.
Although we are aware of the limited number of cases investigated, our findings support the urgent
need to apply the approach presented here for massive search of SARS-CoV-2 in publicly available
human metagenomic databases, currently including hundreds of thousands of human population
metagenomes worldwide (e.g. over 5,000 human faecal metagenomes sequenced on Illumina HiSeq
or NextSeq platform are deposited in the National Center for Biotechnology Information Sequence
Read Archive (NCBI SRA), as accessed on March 18, 2020), as a unique opportunity to retrospectively
map the spread of SARS-CoV-2 in asymptomatic individuals. Our results acquire particular
importance in the context of the very recent publication of Andersen et al. (2020), according to
which it is possible that the progenitor of SARS-CoV-2 jumped into humans from an animal source,
acquiring specific genomic features that provide high affinity for the human receptor ACE2 during
the cryptic human-to-human transmission phase. Once acquired, these adaptations would enable
Electronic copy available at: https://ssrn.com/abstract=3557962
the pandemic to take off and trigger detection by the surveillance system. According to the authors,
this scenario presumes an unrecognized transmission period in humans between the initial zoonotic
event and the subsequent acquisition of the polybasic cleavage site, allowing for greater affinity for
human ACE2. If confirmed on a larger population, our findings may reveal glimpses to support this
hypothesis, providing evidence of the presence of the most recent common ancestor of SARS-CoV-
2 in the human population before the outbreak of the current pandemic, coronavirus disease 2019
(COVID-19), possibly in an inactive non-virulent form, with important implications for the current
unprecedented coordinated international effort to control the global spread of the virus.
References
Andersen, K.G., Rambaut, A., Lipkin, W.I., Holmes E.C., and Garry, R.F. (2020). The proximal origin of
SARS-CoV-2. Nat. Med. doi: 10.1038/s41591-020-0820-9.
Deng, X., Hackbart, M., Mettelman, R.C., O'Brien, A., Mielech, A.M., Yi, G., Kao, C.C., and Baker, S.C.
(2017). Coronavirus nonstructural protein 15 mediates evasion of dsRNA sensors and limits
apoptosis in macrophages. Proc. Natl Acad. Sci. U S A. 114, E4251-E4260. doi:
10.1073/pnas.1618310114.
Hagemeijer, M.C., Rottier, P.J., and de Haan, C.A. (2012). Biogenesis and dynamics of the
coronavirus replicative structures. Viruses 4, 3245–3269. doi:10.3390/v4113245.
Ianiro, G., Mullish, B.H., Kelly, C.R., Sokol, H., Kassam, Z., Ng, S., Fischer, M., Allegretti, J.R., Masucci,
L., Zhang, F., et al. (2020). Screening of faecal microbiota transplant donors during the COVID-19
outbreak: suggestions for urgent updates from an international expert panel. Lancet Gastroenterol.
Hepatol. doi: 10.1016/S2468-1253(20)30082-0.
Knoops, K., Kikkert, M., Worm, S.H., Zevenhoven-Dobbe, J.C., van der Meer, Y., Koster, A.J.,
Mommaas, A.M., and Snijder, E.J. (2008). SARS-coronavirus replication is supported by a
reticulovesicular network of modified endoplasmic reticulum. PLoS Biol. 6, e226.
doi:10.1371/journal.pbio.0060226.
Mas-Lloret, J., Obón-Santacana, M., Ibáñez-Sanz, G., Guinó, E., Pato, M.L., Rodriguez-Moranta, F.,
Mata, A., García-Rodríguez, A., Moreno, V., and Pimenoff, V.N. (2020). Gut microbiome diversity
detected by high-coverage 16S and shotgun sequencing of paired stool and colon sample. Sci. Data
7, 92. doi: 10.1038/s41597-020-0427-5.
Electronic copy available at: https://ssrn.com/abstract=3557962
Nakatsu, G., Zhou, H., Wu, W., Wong, S.H., Coker, O.O., Dai, Z., Li, X., Szeto, C.H., Sugimura, N., Lam,
T.Y., et al. (2018). Alterations in enteric virome are associated with colorectal cancer and survival
outcomes. Gastroenterology 155, 529–541.e5. doi: 10.1053/j.gastro.2018.04.018.
Rampelli, S., Schnorr, S.L., Consolandi, C., Turroni, S., Severgnini, M., Peano, C., Brigidi, P.,
Crittenden, A.N., Henry, A.G., and Candela, M. (2015). Metagenome sequencing of the Hadza
hunter-gatherer gut microbiota. Curr. Biol. 25, 1682-93. doi: 10.1016/j.cub.2015.04.055.
Rampelli, S., Soverini, M., Turroni, S., Quercia, S., Biagi, E., Brigidi, P., and Candela, M. (2016).
ViromeScan: a new tool for metagenomic viral community profiling. BMC Genomics 17, 165. doi:
10.1186/s12864-016-2446-3.
Rampelli, S., Turroni, S., Schnorr, S.L., Soverini, M., Quercia, S., Barone, M., Castagnetti, A., Biagi, E.,
Gallinella, G., Brigidi, P., et al. (2017). Characterization of the human DNA gut virome across
populations with different subsistence strategies and geographical origin. Environ. Microbiol. 19,
4728-4735. doi: 10.1111/1462-2920.13938.
Wang, Q., Zhao, H., Sun, Q., Li, X., Chen, J., Wang, Z., Ju, Y., Jie, Z., Guo, R., Liang, Y., et al. (2019).
Shotgun metagenomics of 361 elderly women reveals gut microbiome change in bone mass loss.
bioRxiv. doi: 10.1101/679985.
Xiao, F., Tang, M., Zheng, X., Liu, Y., Li, X., and Shan, H. (2020). Evidence for gastrointestinal infection
of SARS-CoV-2. Gastroenterology doi: 10.1053/j.gastro.2020.02.055. Online ahead of print.
Xu, Y., Li, X., Zhu, B., Liang, H., Fang, C., Gong, Y., Guo, Q., Sun, X., Zhao, D., Shen, J., et al. (2020).
Characteristics of pediatric SARS-CoV-2 infection and potential evidence for persistent fecal viral
shedding. Nat. Med. doi: 10.1038/s41591-020-0817-4. Online ahead of print.
Figure legends
Figure 1. Workflow of the ViromeScan pipeline for detecting SARS-CoV-2 sequences within
shotgun metagenomics data. Inputs are single-end or paired-end reads. Viral reads are identified
using ViromeScan and a reference database containing DNA and RNA viruses, including the SARS-
CoV-2 genome. The viral sequences assigned to the Coronaviridae family are finally mapped to
another database containing 100 sequences of coronavirus genomes.
Figure 2. Virome profiling and detection of SARS-CoV-2 traces in 38 human faecal metagenomes.
Hierarchical Ward-linkage clustering based on Spearman correlation coefficients of family-level viral
profiles of 38 human faecal metagenomes from different studies, with different geographic origin
Electronic copy available at: https://ssrn.com/abstract=3557962
and sampling window (26 Chinese elderly women sampled before April 2019 (Wang et al., 2019); 3
Italian adults sampled in March-April 2013 (Rampelli et al., 2015); 9 Spanish adults sampled before
November 2019 (Mas-Lloret et al., 2020)), as determined by ViromeScan with the additional COVID-
19 module. The analysis was carried out considering all the families detected with at least 0.1 counts
per million (CPM) of abundance in 3 samples. The dot column on the right refers to the
presence/absence of reads for SARS-CoV-2.
Electronic copy available at: https://ssrn.com/abstract=3557962
ATTCGCTGAACTATTTTAAGGC
TAGTAGCCAATACGTACCCTAGTTAATGTAG
Metagenome
Sequencing
CCTGAGATCACTATTAAG
VIROMESCAN suite
• Alignment against human viral database
implemented with SARS-CoV-2 sequence
• Filtering of human sequences
• Filtering of bacterial sequences
New COVID-19 module
• Realignment of hits for Coronaviridae against the
Coronavirus database
Reads for SARS-CoV-2
Electronic copy available at: https://ssrn.com/abstract=3557962
Wang et al. 2019 Coronaviridae detected
Rampelli et al. 2015 Coronaviridae detected and SARS-CoV-2 detected
Mas-Lloret et al. 2020 Coronaviridae NOT detected
Total_mean
0 30 CPM 200
SRR8845291
SRR8845291 SRR8845291
SRR8845292
SRR8845293
SRR8845292 SRR8845292
SRR8845294
SRR8845295
SRR8845296
SRR8845293 SRR8845293
SRR8845297
SRR8845298
SRR8845294 SRR8845294
SRR8845299
SRR8845300
SRR8845295 SRR8845295
SRR8845301
SRR8845302
SRR8845303
SRR8845296 SRR8845296
SRR8845304
SRR8845305
SRR8845297 SRR8845297
SRR8845306
SRR8845307
SRR8845298 SRR8845308
SRR8845298
SRR8845309
c(38:1)
20
SRR8845310
30
SRR8845299 SRR8845299
SRR8845311
SRR8845312
SRR8845300 SRR8845300
SRR8845313
SRR8845314
SRR8845315
SRR8845301 SRR8845301
SRR8845316
IT2
SRR8845302 SRR8845302
IT5
IT6
SRR8845303 SRR8845303
ERR3450203
ERR3450229
ERR3450296
SRR8845304 SRR8845304
ERR3450606
ERR3451635
SRR8845305 SRR8845305
ERR3452318
ERR3452529
SRR8845306 SRR8845306
ERR3452574
ERR3452699
SRR8845307 SRR8845307
Anelloviridae
Orthomyxoviridae
Arenaviridae
Papil omaviridae
Bunyaviridae
Herpesviridae
Poxviridae
Hepeviridae
Coronaviridae
Flaviviridae
SRR8845308 SRR8845308
c(38:1)
10
20
SRR8845309 SRR8845309
SRR8845310 SRR8845310
SRR8845311 SRR8845311
SRR8845312 SRR8845312
SRR8845313 SRR8845313
SRR8845314 SRR8845314
SRR8845315 SRR8845315
SRR8845316 SRR8845316
IT2 IT2
IT5 IT5
10
IT6 IT6
0
ERR3450203 ERR3450203
ERR34502290.6 0.8 1.0 1.2 1.4
ERR3450229
ERR3450296 ERR3450296
rep(1, 38)
ERR3450606 ERR3450606
ERR3451635 ERR3451635
ERR3452318 ERR3452318
ERR3452529 ERR3452529
ERR3452574 ERR3452574
ERR3452699 ERR3452699
0
Anelloviridae
Orthomyxoviridae
Arenaviridae
Papillomaviridae
Bunyaviridae
Herpesviridae
Poxviridae
Hepeviridae
Coronaviridae
Flaviviridae
SARS-CoV-2
0.6 0.8 1.0
rep(1, 38)
Electronic copy available at: https://ssrn.com/abstract=3557962
Supplemental Information
Experimental procedures
Metagenomic sequences were downloaded from the ENA and SRA repositories using the following
project IDs: PRJNA278393 (Rampelli et al., 2015), PRJNA530339 (Wang et al., 2019), and
PRJEB33098 (Mas-Lloret et al.. 2020). Only deep sequenced samples (> 10 M reads) were used for
the analysis.
In order to detect traces of the SARS-CoV-2 genome within the data, we adopted the ViromeScan
pipeline (Rampelli et al., 2016), with a new COVID-19 module, released together with this study
(https://github.com/simonerampelli/viromescan). In short, input single-end and paired-end reads
in fastq format are aligned to the human DNA/RNA virus database included in ViromeScan and
implemented with the reference sequence for SARS-CoV-2 (NCBI ID: NC_045512.2). The database
contains only viruses that have the human being as a natural host and is based on the complete viral
genomes available on the NCBI website. The first alignment step is performed using bowtie2 and is
necessary for a complete and accurate screening of the sequences to select viral reads. Afterwards,
the selected reads are filtered by quality and human contamination following the procedure of the
Human Microbiome Project (Turnbaugh et al., 2007). Human-filtered reads are also screened for
bacterial cross-matching, to avoid biases due to bacterial contamination. At this point, filtered reads
are again compared with the viral genomes of the database using bowtie2, allowing the definitive
assignment of each virome sequence to a viral genome. Finally, reads assigned to Coronaviridae are
retrieved and compared to another database containing 45 genome sequences for SARS-CoV-2 and
55 genomes for other coronaviruses, also from different hosts. This final comparative alignment is
necessary to increase the discriminating power between SARS-CoV-2 and related genomes. All the
data, instructions and script sources needed to run the pipeline are available on the ViromeScan
website provided above. For each sample analysed, the total amount of counts was normalized to
CPM (counts per million of sequences). CPM <0.1 were considered below the detection limit and
set to 0 for all samples. The presence of SARS-CoV-2 was assessed in each metagenome by
considering the absolute number of reads assigned to SARS-CoV-2 and the relative contribution of
these reads to the total number assigned to Coronaviridae. Specifically, more than 3 counts and
more than 30% respectively, were considered as proof of presence. The heatmap was built using
the R package “gplots” (https://github.com/talgalili/gplots).
Electronic copy available at: https://ssrn.com/abstract=3557962
Supplementary references
Mas-Lloret, J., Obón-Santacana, M., Ibáñez-Sanz, G., Guinó, E., Pato, M.L., Rodriguez-Moranta, F.,
Mata, A., García-Rodríguez, A., Moreno, V., and Pimenoff, V.N. (2020). Gut microbiome diversity
detected by high-coverage 16S and shotgun sequencing of paired stool and colon sample. Sci. Data
7, 92. doi: 10.1038/s41597-020-0427-5.
Rampelli, S., Schnorr, S.L., Consolandi, C., Turroni, S., Severgnini, M., Peano, C., Brigidi, P.,
Crittenden, A.N., Henry, A.G., and Candela, M. (2015). Metagenome sequencing of the Hadza
hunter-gatherer gut microbiota. Curr. Biol. 25, 1682-93. doi: 10.1016/j.cub.2015.04.055.
Turnbaugh, P.J., Ley, R.E., Hamady, M., Fraser-Liggett, C.M., Knight, R., and Gordon, J.I. (2007). The
human microbiome project. Nature 449, 804–10. doi: doi.org/10.1038/nature06244.
Wang, Q., Zhao, H., Sun, Q., Li, X., Chen, J., Wang, Z., Ju, Y., Jie, Z., Guo, R., Liang, Y., et al. (2019).
Shotgun metagenomics of 361 elderly women reveals gut microbiome change in bone mass loss.
bioRxiv. doi: 10.1101/679985.
Electronic copy available at: https://ssrn.com/abstract=3557962
