Integrated graphical analysis of protein sequence features predicted from sequence composition

Several protein sequence analysis algorithms are based on properties of ami no acid composition and repetitiveness. These include methods for predictio n of secondary structure elements, coiled-coils, transmembrane segments or signal peptides, and for assignment of low-complexity, nonglobular, or intr insically unstructured regions. The quality of such analyses can be greatly enhanced by graphical software tools that present predicted sequence featu res together in context and allow judgment to be focused simultaneously on several different types of supporting information. For these purposes, we d escribe the SFINX package, which allows many different sets of segmental or continuous-curve sequence feature data, generated by individual external p rograms, to be viewed in combination alongside a sequence dot-plot or a mul tiple alignment of database matches. The implementation is currently based on extensions to the graphical viewers Dotter and Blixem and scripts that c onvert data from external programs to a simple generic data definition form at called SFS. We describe applications in which dot-plots and flanking dat abase matches provide valuable contextual information for analyses based on compositional and repetitive sequence features. The system is also useful for comparing results from algorithms run with a range of parameters to det ermine appropriate values for defaults or cutoffs for large-scale genomic a nalyses. (C) 2001 Wiley-Liss, Inc.