9 A Pair Ranking (PRank) Method for Assessing Assay …
175
9.5.5 Toward Biological Data-Based Read-Across
Read-across is a data gap filling strategy to assess the toxicity of untested compounds
based on the analog to similar chemicals, for which toxicity data are available. Readacross approaches have been widely adopted by regulatory programs such as OECD
and REACH [52]. Besides structure and physicochemical properties, the biological
data-based similarity may enhance this process [53].
We compared compound similarity between chemical space and three preclinical TGx assays (Fig. 9.8). We observed low Pearson’s correlation coefficients of
0.30, 0.20, and 0.21 for chemical space versus InVitro, InVivo_S, and InVivo_R,
respectively. The divergence between chemical space and toxicogenomic space indicated that toxicogenomics data might provide extra value to improve chemical-based
Read-across.
Fig. 9.8 Concordance between chemical space and TGx space: the compound pairwise chemical
similarity was calculated based on ECFP-4 fingerprints by using Pipeline Pilot. Pearson’s correlation
coefficients were calculated between chemical space and three TGx assays, respectively
Précédent

- 186/416

Suivant