Networks of information token recurrences derived from genomic sequences may reveal hidden patterns in epidemic outbreaks: A case study of the 2019-nCoV coronavirus.

Markus Luczak-Roesch

Inter-and intra-cluster similarity From our measurement of the intra-cluster similarity (see Table 1 ) we find that the genomic similarity at the level of nucleotides is low in the central cluster 0, which is expected because this is the largest of all clusters with low temporal coherence. However, cluster 5 also features a slightly lower average similarity with comparably high standard deviation, which indicates that this cluster may be slightly less coherent. The barchart of the inter-cluster analysis results shown in Figure 5 reveals that there are some cluster pairs with high inter-cluster similarity, that the inter-cluster similarity is different between the +3 reading frame network and the other two reading frame networks, and that the pattern of the +3 reading frame network inter-cluster similarity is similar to the one for the entire TIC network. In particular we find that in the +1 and +2 reading frame networks the clusters 1, 4, 6 and 7 are quite similar at the nucleotide level compared to all other clusters that are pairwise distinct. For the entire TIC network and the +3 reading frame network instead we find that clusters 1, 3, 5 and 7 show this kind of similarity. Altogether, these results indicate that (a) in some instances the clustering based on structural characteristics of the TIC network overrides actual genomic similarity, and (b) the choice of the reading frames during TIC construction has an impact on the resulting network that needs to be further investigated.