Fig. 2From: Evaluation of BLAST-based edge-weighting metrics used for homology inference with the Markov Clustering algorithmIllustration of simulated sequence fragmentation. This example illustrates the four different ways in which the fragmentation scheme 1323 would be applied to a toy input test database with only three clusters containing sequences from only four organisms. The four resulting test sets represent a cross of two variables, arranged here into rows and columns. The sequences in the top row (a & b) have been split into even subsequences. The sequences in the lower row (c & d) have been randomly fragmented into uneven subsequences. In the left column (a & c), the user-defined integer assigned to each organism directly determines the number of subsequences into which each sequence is split. In the right column (b & d), these integers are first mapped to all sequences within a cluster, but are then shuffled within that cluster before fragmentationBack to article page