Fig. 1
From: Cnidaria: fast, reference-free clustering of raw and assembled genome and transcriptome NGS data
Cnidaria analysis summary. The JELLYFISH software reads each of the source sequence files (in Fasta or Fastq formats), extracts their k-mers (k = 3 in this example), canonizes them (by generating the reverse complement of each k-mer and storing only the k-mer which appears first lexicographically), orders them according to a deterministic hashing algorithm (in this example, alphabetically) and then saves each dataset in a separated database file (.jf). CNIDARIA subsequently reads these databases and compares them, side-by-side, by counting the total number of k-mers (white circles), the number of valid k-mers (k-mers shared by at least two samples, black circles) and the number of shared k-mers for each pair of samples as a matrix. Those values are exported to a Cnidaria Summary Database (CSD, a .json file) that is then used to construct a matrix of, by default Jaccard, distances between the samples (Formula 1). This dissimilarity matrix is then used for Neighbour-Joining clustering and exported as a NEWICK tree. Alternatively, Cnidaria can export a Cnidaria Complete Database (CCD, a .cne file) containing all k-mers and a linked list describing their presence/absence in the samples. This second database can be used as an input dataset together with other .cne or .jf files for new analysis