Discovery: Integrative Precision Medicine Tool
There are many sophisticated approaches described in literature that find signatures or integrate multiple data types to predict drug response or patient prognosis. In this poster we describe the research, testing, and evaluation of the results from one such tool to use within the Translational Genomics Research Institute.
The tool that best met our established criteria was PharmacoGX and this was tested using the internal IVY dataset containing molecular profiles on patients with Glioblastoma. The results were compared to drug rules linking molecular alterations to drug recommendations based on manual curation of evidence in the scientific literature. The dataset was comprised of data from 13 separate patients. We used both mutation and RNA data from IVY. We tested two functions - perturbation/connectivity and sensitivity modeling. Perturbation studies look at the molecular profiles of a cell line before and after the application of a drug and can characterize the effect of the drug. The perturbation signatures created by the tool were compared to the IVY gene expression data using connectivity scoring. These connectivity scores are used to find candidate drugs for reversing the disease signature. Sensitivity modeling allows the discovery of molecular features that are correlated with sensitivity of cell lines to compounds. This type of analysis can be used to select drugs that are tailored to an individual patient using the individual patients gene mutations as an input.
Overall, the sensitivity modeling results were inconclusive, with most of the results not being upheld by literature, a separate tool called CellMinerCDB, or the internal glioblastoma curated drug rules. The connectivity scoring appeared promising. We looked at all significant drugs that were also found within the internal glioblastoma curated drug rules. We also reviewed the results that were found to be significant for every patient in the study. While none of these results were included in the IVY database, the results were supported by outside literature. In particular, two drugs (Chlorogenic Acid and Indometacin) were predicted to reverse the tumor signatures of the patients and act as anti-tumor drugs for glioblastoma. Outside literature supported both findings. These two drugs warrant further investigation and potential testing in the lab.