Leaders in Pharmaceutical Business Intelligence (LPBI) Group

Predict Toxicity from Structure

Larry H. Bernstein, MD, FCAP, Curator

LPBI

 

chemTox

http://www.cyprotex.com/insilico/physiological_modelling/chemtox

Toxicity Prediction Directly from Chemical Structure

• Predicts key toxicity parameters (Ames mutagenicity, rat acute dose LD₅₀ following iv or po administration) and aqueous solubility directly from structure.
• No in vitro physicochemical or toxicity data required.
• Save money and time by allowing toxicity to be assessed virtually (no synthesis required).
• Provides early stage filter for directing chemistry and prioritizing screening.

A virtual screening tool for predicting toxicity from chemical structure alone.

 

chemTox Input Requirements	Chemical structure, e.g. SMILES, mol or sdf
chemTox Data Delivery	Predicted toxicity measures: Ames mutagenicity, rat acute dose LD50 following iv or po administration Predicted aqueous solubility

 

chemTox is implemented as a node for the KNIME analytics platform which executes a model of the workflow illustrated in Figure 1 below.

Figure 1
chemTox workflow.

KNIME can be downloaded easily and for free. Cyprotex then provide the bespoke toxicity modules (chemTox) for the KNIME platform.

Properties

The following properties are reported: Ames mutagenicity classification (mutagenic/non-mutagenic). Ames mutagenicity probability (probability of being a mutagen). Rat LD50 (mmol/kg) following acute administration by intravenous route. Rat LD50 (mmol/kg) following acute administration by oral route. Aqueous solubility (mol/l). Output is a KNIME data table facilitating saving to a file or database, or using the predictions as inputs to subsequent workflow steps.

Model Development

Models are quantitative-structure property relationships (QSPR) that calculate the toxicity properties of interest in terms of a compound’s structural descriptors. Models have been trained using large, well-validated datasets. Training was performed using up-to-date, rigorous statistical pattern recognition methods. Repeated 10-fold cross-validation has been used to generate the most robust statistics for prediction performance.

Data from chemTox

Figure 2
Receiver operating characteristic (ROC) plots for classification of rat acute LD₅₀ following oral administration. (a) LD₅₀ < 5mg/kg (b). LD₅₀ < 50mg/kg (c). LD₅₀ < 300mg/kg.

(a).
LD₅₀ < 5mg/kg

(b).
LD₅₀ < 50mg/kg

(c).
LD₅₀ < 300mg/kg

Statistics
Area under the ROC curve	0.905
Sensitivity	0.85
Specificity	0.82

Table 1
Statistics for predicted Ames mutagenicity based on 4336 compounds (1935 non-mutagenic and 2401 mutagenic).

Model variants can be generated having different balance of sensitivity and specificity to suit different screening requirements.

 

SJ Williams

Tempting but there have been other databases like ToxNet which have used SAR (structure activity relationship) for Ames prediction but usually that is far as it can go. This is usually a first screen but long term carcinogenicity studies are still required for obvious reasons. In addition no system can predict IDILI (idiopathic drug induced liver injury) which has and still is plaguing the drug development industry. This is where relationships need to be worked on and this problem will need a more indepth ‘omic strategy as there are no SARs which correlate with these “idiosyncratic” toxicities. However be that as it may there have been chemicals which failed the Ames test however were not carcinogenic in subchronic or chronic tests so it is still good to conduct those studies anyway, even though most regulatory bodies give a failed Ames test a bad thumbs down.

Share this:

• Tweet
• Print
• Email
• More

Like this:

Like Loading...

Posted in Drug Toxicity | Tagged Area Under Curve (AUC), Toxicity Prediction | 1 Comment

One Response

1. on November 17, 2015 at 5:21 PM | Reply sjwilliamspa

   This is very insightful. There is no doubt that there is the bias you refer to. 42 years ago, when I was postdocing in biochemistry/enzymology before completing my residency in pathology, I knew that there were very influential mambers of the faculty, who also had large programs, and attracted exceptional students. My mentor, it was said (although he was a great writer), could draft a project on toilet paper and call the NIH. It can’t be true, but it was a time in our history preceding a great explosion. It is bizarre for me to read now about eNOS and iNOS, and about CaMKII-á, â, ã, ä – isoenzymes. They were overlooked during the search for the genome, so intermediary metabolism took a back seat. But the work on protein conformation, and on the mechanism of action of enzymes and ligand and coenzyme was just out there, and became more important with the research on signaling pathways. The work on the mechanism of pyridine nucleotide isoenzymes preceded the work by Burton Sobel on the MB isoenzyme in heart. The Vietnam War cut into the funding, and it has actually declined linearly since.

   A few years later, I was an Associate Professor at a new Medical School and I submitted a proposal that was reviewed by the Chairman of Pharmacology, who was a former Director of NSF. He thought it was good enough. I was a pathologist and it went to a Biochemistry Review Committee. It was approved, but not funded. The verdict was that I would not be able to carry out the studies needed, and they would have approached it differently. A thousand young investigators are out there now with similar letters. I was told that the Department Chairmen have to build up their faculty. It’s harder now than then. So I filed for and received 3 patents based on my work at the suggestion of my brother-in-law. When I took it to Boehringer-Mannheim, they were actually clueless.

   Loading...