DGraph

DGraph visualizes protein sequence relationships by generating alignment-free 2-dimensional distance graphs that cluster sequences using physico-chemical property (PCP)–based similarity and Property Distance (PD) scores.

Key Features:

• Alignment-Free Clustering: Employs a dynamic programming approach to generate 2D maps from similarity scores derived from physico-chemical property (PCP) similarities, enabling analysis of unaligned protein sequence data.
• Property Distance Scores: Calculates Property Distance (PD) scores that quantify interrelatedness of sequences based on PCPs and displays these scores graphically.
• Biological Consistency: Produces PD-graphs validated across flaviviruses (FV), enteroviruses (EV), and coronaviruses (CoV) with cluster patterns reflecting vector types, host species, cellular receptors, and disease phenotypes.
• Evolutionary Insights: Reveals connectivities and potential evolutionary relationships that may be challenging to discern with alignment-based methods such as bootstrapping.

Scientific Applications:

• Vector-type differentiation: Differentiates tick-borne and mosquito-borne flaviviruses based on PCP-derived clustering.
• Host-species clustering: Clusters viruses by host species (e.g., bats, camels, seabirds, humans) and correlates these clusters with disease phenotypes.
• Coronavirus spike protein segregation: Segregates beta-coronavirus spike proteins of SARS, SARS-CoV-2, and MERS from other human pathogenic coronaviruses in patterns consistent with cellular receptor usage.

Methodology:

Computes PD scores from unaligned FASTA files using a dynamic programming algorithm based on physico-chemical property similarities and projects those scores into 2D distance maps.