Abstract

Sir, The COVID-19 pandemic has stalled the world and catapulted the global health systems into unprecedented chaos. More than 200 countries have been affected by this pandemic, resulting in 2.54 million cases in a short period of time and >0.17 million deaths (as of April 23, 2020), with a mere 0.7 million recoveries1. The movement of COVID-19 hotspot from China to Europe, and now to the USA, has been partly due to the staggered restrictions in global travel and partly due to potent transmission through asymptomatic carriers2. India, with 21,393 cases and 681 deaths (as of April 23, 2020)1, had the lowest figures for any country of the comparable population (0.5 deaths per million population). International travellers or their close contacts formed the majority of initially reported cases. The delayed onset of COVID-19 in India has given it an edge, which allowed it to impose severe restrictions to contain the local spread 345. In our in-depth analyses of 1500+ genomes, variability among clinical isolates was shown along the timeline, leading to distinct clustering of SARS-CoV-2 across the globe (unpublished observation). It was predicted, based on the aggregation propensity of the spike protein in the Wuhan and other isolates of SARS-CoV-2, that this virus would exhibit very high transmissibility and confer survival fitness67. Genetic diversity of the virus increases with disease progression and can be utilized to model the evolution and propagation of the disease6 Recently, phylogenetic network analysis of 160 SARS-CoV-2 genome samples showed a parallel evolution of the virus and its evolutionary selection in their human hosts8. Similar whole-genome analyses of the Indian isolates and their comparison with global isolates can provide a better understanding of dominant clades within the population and unveil targets for developing specific interventions. In the present study, machine learning-based t-SNE analysis of global clinical isolates has been utilized to segregate the clinical isolates into clusters while accommodating the outliers910. Whole-genome analysis of 3968 global isolates obtained from GISAID (Global initiative on sharing all influenza data)11, including 25 SARS-CoV-2 genomes sequenced in India [next-genome sequencing (NGS) data submitted by the ICMR-National Institute of Virology, Pune, India] and presented in (Figure 1) (Supplementary Fig. 1 (available from http://www.ijmr.org.in/articles/2020/151/5/images/IndianJMedRes_2020_151_5_474_284485_sm5.pdf) and [Supplementary Table 1 (available from http://www.ijmr.org.in/articles/2020/151/5/images/IndianJMedRes_2020_151_5_474_284485_sm6.pdf)), was an attempt to dissect the global genome diversity and also critically evaluate the placement of Indian isolates to understand the COVID-19 pandemic in India.Fig. 1: Whole-genome-based t-SNE clustering of 3968 clinical isolates. (A) Comparative genome-based clustering of Indian isolates (red) with Chinease isolates (blue). (B) Comapartive genome-based clustering of Indian isolates (red) and Chinese isolates (blue) with rest of the world (green). (C) Diversity in clinical isolates showing three distinct clustering using hierarchical clustering on the t-SNE clusters. (t-SNE: https://github.com/jdonaldson/rtsne).Supplementary Table I: Details of Indian samples along with their origin and placement in Hierarchical clusters. Genomic sequences were retrieved from GISAID (https://www.gisaid.org)The initial cases reported from India had a travel history to China, which explained its position in a Chinese cluster5 (Fig. 2). The travel ban from China to India, in early February 2020, has prevented the large-scale spill-over directly from China to the Indian Sub-continent. However, various isolates transmitted from other South-East Asian countries might fall in the same cluster. The overlap of Indian samples majorly with European samples (Supplementary Fig. 1, Panel III) reiterated the fact that the delayed travel restriction from the European hotspot regions affected not just India but also many countries.Fig. 2: Position of various Indian isolates with other nations. (A-E) Clustering of SARS-CoV-2 genome sequences from India (red) with other nations around the globe. Indian samples clustered with samples from different nations – China, Kuwait, Canada, USA and Spain in whole-genome-based clustering. Figures were generated using FigTree v1.4.4 (http://tree.bio.ed.ac.uk/software/figtree/).Hierarchical-based clustering further yielded exciting outcomes on the inter-continent transmission of COVID-19. The segregation of SARS-CoV-2 genomes into three clades indicates the emergence of evolutionary diversity (Fig. 1C). The heterogeneity of these clusters, grouped along with Chinese counterparts, validates a global spill-over event originating from Wuhan512. Hierarchical cluster 2 in Supplementary Figure 1 Panel II (coloured by the continents) indicates the introduction of SARS-CoV-2 in India from the European, other Asian and North American nations (Supplementary Fig. 1). Detailed comparative analysis of Indian isolates with respect to other countries showed its close relationship with samples from China, USA, Canada, Spain and Kuwait, suggestive of exposure to COVID-19 due to travel history from these nations (Fig. 2). However, limited genome sequences from India make it difficult to differentiate and ascertain global transmission and transmission within the country. The conservation of an amino acid in any protein sequence denotes its functional importance1314 as it undergoes fewer amino acid replacements or is more likely to substitute amino acids with similar biochemical properties. The amino acid conservation is inversely proportionate to the evolutionary rate. This is a valuable gauge of the evolutionary divergence and the analogous genomic regions. Sequence similarity between the open reading frames (ORFs) of Indian isolates and the initial sample collected in Wuhan unravels conservation in five ORFs corresponding to envelope protein, membrane glycoprotein, ORF6, ORF7b and ORF10 proteins (Fig. 3A). On the contrary, a number of mutations were observed in ORF1a, ORF1b, spike protein (surface glycoprotein), ORF3a, ORF7a, ORF8 and nucleocapsid phosphoprotein (Supplementary Fig. 2 (available from http://www.ijmr.org.in/articles/2020/151/5/images/IndianJMedRes_2020_151_5_474_284485_sm7.pdf and Supplementary Table II (available from http://www.ijmr.org.in/articles/2020/151/5/images/IndianJMedRes_2020_151_5_474_284485_sm8.pdf)). Mean similarity calculated for these ORFs revealed that ORF1a in the Indian isolates was less conserved (more mutated) compared to global isolates (Fig. 3A and Supplementary Table III (available from http://www.ijmr.org.in/articles/2020/151/5/images/IndianJMedRes_2020_151_5_474_284485_sm9.pdf)). In all other ORFs, a relatively higher conservation was observed among Indian isolates compared to Wuhan strain. When compared with global isolates, Indian isolates have higher entropy for changes in ORF 1a and ORF 1b (Supplementary Fig. 3 (available from http://www.ijmr.org.in/articles/2020/151/5/images/IndianJMedRes_2020_151_5_474_284485_sm10.pdf)).. Further, qualitative analysis of mutations in non-conserved ORFs showed that each type of amino acid had undergone mutation in the Indian isolates (Fig. 3B). These mutations could be a major contributing factor for the separation of Indian isolates into three distinct clusters. Higher sampling rate driven by NGS of the Indian isolates would help in better understanding of actual variability in SARS-CoV-2 and assist both in identifying better diagnostic markers and in developing specific interventions in terms of vaccine candidates and drug targets.Fig. 3: Sequence similarity and mutation analysis of open reading frames. (A) Comparison of mean sequence similarity for open reading frames between Indian and global isolates with Wuhan strain. (B) Qualitative analysis on type of mutations occurring in non-conserved open reading frames (ORFs) of Indian isolates compared to Wuhan strain.Supplementary Table II: Specific high frequency (>=10%) mutations in individual ORFs in Indian isolates compared with reference strain (Wuhan_IPBCAMS-WH-01_2019_EPI_ISL_402123). The sequences with less than 25% gaps were selected for all the studies. Genomic sequences were retrieved from GISAID (https://www.gisaid.org)Supplementary Table III: Comparison of mean sequence similarity for ORFs between Indian and global isolates with Wuhan strainEvolutionary divergence, corroborated by epidemiological data, is a valuable tool to implement appropriate measures against this pandemic. The population density of India and the presence of functionally distinct isolates in the Indian population raise concerns and warrant an urgent need for higher sampling rate for better assessment of the evolution of SARS-CoV-2 in India. The situation is further confounded by the fact that many of these Indian isolates submitted in databanks include those of Indians living in Iran, Italian tourists visiting India, and also contains samples cultured in vitro. In conclusion, a whole-genome diversity analysis of 3968 global clinical isolates, including 25 isolates sequenced in India, of SARS-CoV-2 was done. The variations in different open reading frames (ORFs) of SARS-CoV-2, which drives the formation of distinct Indian clusters and functional heterogeneity, were highlighted. Five ORFs corresponding to envelope protein, membrane glycoprotein, ORF6, ORF7b and ORF10 were found to be highly conserved, while a number of mutations were observed in ORF1a, ORF1b, spike protein, ORF3a, ORF7a, ORF8 and nucleocapsid phosphoprotein. Generating diverse genomic datasets will provide insight into the propagation dynamics of COVID-19, leading to a better understanding of pathogenesis and evolution of SARS-CoV-2, which will eventually lead to better intervention methods. Acknowledgment The seventh author (SEH) acknowledges Department of Biotechnology, Government of India for funding support (BT/PR23099/NER/95/632/2017), (BT/PR23155/NER/95/634/2017). SEH is a JC Bose National Fellow, Department of Science and Technology, Government of India & Robert Koch Fellow, Robert Koch Institute, Berlin. The first author (HS) is a recipient of Women Scientist fellowship, Department of Health Research and the second (JS) & fourth (SJ) authors received Young Scientist fellowships from the Department of Health Research, Ministry of Health and Family Welfare, Government of India. The sixth author (JAS) received UGC Startup grant and the third author (MK) received Silver Jubilee Post-Doctoral fellowship from Jamia Hamdard, New Delhi. Authors acknowledge the Originating and Submitting Laboratories for their sequences and meta-data shared through GISAID on which this study is based. Authors acknowledge BioInception Pvt. Ltd, for providing their proprietary data analysis pipeline and platform.