Skip to main content
Fig. 5 | BMC Bioinformatics

Fig. 5

From: Evaluation of BLAST-based edge-weighting metrics used for homology inference with the Markov Clustering algorithm

Fig. 5

Sensitivity performance comparison for each edge-weighting metric and fragmentation scenario. Sensitivity performance of MCL on graphs weighted using each of the four metrics (columns) over a range of inflation parameter values (x-axes) in a variety of fragmentation scenarios (rows). Vertically stacked bars indicate the number of clusters into which the members of a particular ECK have been split. Blue segments represent ECKs that were completely contained within a single MCL cluster. Other segments represent ECKs that were split into two or more MCL clusters, with redder color indicating higher degrees of fragmentation. The number of ECKs is fixed, so each stack sums to exactly 458. Five different simulated fragmentation scenarios are displayed as faceted rows: 11111111111 – all sequences are intact; 11122222112 – approximately half of the sequences have been split into two pieces; 22222222222 – all sequences have all been split into two pieces; 33311222111 – approximately one third of the sequences have been split into two pieces, one third have been split into three pieces, and the remaining third were left intact; 33333333333 – all sequences were split into three pieces

Back to article page