Table 1.
Parameter Estimates for Nonoverlapping PNGSs with Significant Interactions in the Disequon Model
Figure 1.
The Nature of Glycan Interactions Depends on Their Intramolecular Distance
Potential PNGSs that were identified by the disequon model as having significant interactions with other PNGSs were mapped to a structural model of HIV-1 gp120. Intramolecular distances between interacting PNGSs were measured using the visualization software UCSF Chimera [53]. When an asparagine residue corresponding to a PNGS was not present in the structural model, the distance between pairs was approximated by 4.0 Å times the number of residues separating the PNGSs, to a maximum of five residues. If the number of residues exceeded five, then the unmapped pair was omitted from the analysis.
Figure 2.
Consensus Bayesian Network of Highly Polymorphic PNGSs in the HIV-1 Envelope
(A) Ordered distribution of frequencies that undirected links (i.e., arcs) occurred in replicate networks optimized on 100 random samples of 200 sequences from the alignment, expressed as a percentage (y-axis). A dotted line at 50% indicates an arbitrary cutoff, above which arcs were applied to assembling the consensus network.
(B) Graph of consensus network. Each node (oval) represents the PNGS identified by its position in the HxB2 reference sequence, with one exception (V2ins) that represents an inserted PNGS downstream of the conserved N186. Arcs between nodes are labeled with percentage of occurrence above and odds-ratios below. An odds-ratio <1 (red) indicates mutual exclusion and >1 (blue) co-occurrence of PNGSs.
Table 2.
Posterior Probabilities of Undirected Arcs in Subtype-Specific Networks
Figure 3.
Interacting PNGS Mapped to Structural Model of gp120
Arcs from the consensus network are mapped to asparagine residues in the structure of a folded CD4-bound gp120 protein. The host cell-binding face of gp120 is oriented towards the top of the figure, and variable loops (of which many residues are truncated) are highlighted in pink. Arcs are colored red to indicate mutual exclusion, and blue indicates co-occurrence of the PNGSs. The asparagine residues of mutually exclusive PNGSs N295/N442 and N362/N465 are located within ∼12 Å of one another. In contrast, co-occurent PNGSs are separated by much greater intramolecular distances.
Figure 4.
This diagram illustrates the eight rates that are parameterized by the disequon (paired-character) model. Each box corresponds to a state in the disequon model, in which the first digit represents the presence or absence of a PNGS at the first site, and the second digit for the second site. The interaction parameter (ɛ) between paired sites is highlighted in red.