CpG Relational Graphs
Separating the graphs is intended to prevent semantic mixing between continuous correlation edges and Boolean annotation edges. Co-methylation was found to contribute the strongest relational signal for age estimation. The study constructs…
1 sources - 6 claims
Separating the graphs is intended to prevent semantic mixing between continuous correlation edges and Boolean annotation edges. Co-methylation was found to contribute the strongest relational signal for age estimation. The study constructs co-methylation, same-chromosome, and same-gene graphs over the same CpG node set. The co-methylation graph connects CpG pairs whose methylation correlation exceeds a threshold. Same-chromosome and same-gene graphs use Boolean-positive annotation relationships. The co-methylation graph is sparsified to reduce density, memory cost, and noise.