Research Scientist
Agouron Pharmaceuticals, Inc

Tasks

My work consists of three types of tasks:

1. I do a lot mathematical modeling and simulation, especially of the workings of a cell. I simulate the biochemical kinetics of the process of DNA synthesis. The anti-cancer project I am working on now is trying to inhibit a particular enzyme to shut down a tumor’s ability to make DNA, and thus, to grow. It is very difficult to predict the so-called downstream effects of turning off this particular enzyme for two reasons, namely that there are so many interconnecting, cross-regulating, cyclical biochemical pathways are involved, and that it is impossible to measure experimentally everything we would like to measure. Putting together the model, validating it and then using it in simulations as a predictor helps us understand what is happening inside the cell. When the simulation results do not agree with experimental results, it tells us that we do not understand the cellular processes as well as we think we do. I have used the results from these in silico experiments to provide one possible explanation of surprising in vitro results. I have also done similar modeling on the kinetics of viral replication. That modeling helped the project team refocus its goal for one property of a candidate compound.

2. I analyze experimental data. One of my first tasks at Agouron was doing a mathematical analysis of the available techniques for analyzing in vitro drug combination experiments. (For example, most people with HIV take 3 drugs every day to suppress the virus. Many people with cancer have 2- or 3-drug chemotherapy.) It turns out that the two most popular techniques have unfixable mathematical problems in their derivations. (I think the original reviewers of the journal articles did not have the mathematical skills to find the errors.) A third type of analysis, which uses surface fitting, is too difficult for most of the biologists and doctors who study the effects of drugs in combination. I wrote the tools we use to analyze this type of experiment, and I do most of the analysis for our combination drug experiments. I also help design the experiments so that the resultant data will be analyzable.

3. I provide mathematical and statistical support, from ba sic to sophisticated, to my colleagues in pharmacology and virology. I answer statistical questions, write and install macros and templates to perform calculations, and assess software and statistical techniques for their applicability to the work we do. I have also been brought in to check whether data generated by certain experiments can answer the questions they are supposed to address. In one case this involved the looking at the variability of the data, the detection limits of the equipment for the assay, the growth rates of the cells and the virus being used, and the variability of extrapolations from previous experiments.

Research Scientist
Agouron Pharmaceuticals, Inc

Issues

Below is a copy of some information I put together for a talk at last summer’s Workshop on Mathematics in Industry, hosted by the University of San Diego, about how mathematics can be used in drug development (as opposed to what I do).

Two Relevant Mathematical Issues in Drug Development

I. Extracting information from an enormous amount of data

1. Functional genomics
   1. Which sequences code for proteins and which sequences are junk?
   2. For a sequence that codes for a protein, what does that protein do?
2. Predicting biological properties of a chemical compound in a drug development project
   1. Of the 10s, 100s or 1000s of chemical compounds that inhibit the target enzyme, which ones will
      1. ... be soluble in solvents that animals can tolerate?
      2. ... get into a cell?
      3. ... be orally bioavailable?
      4. ... will not be metabolized to inactive compounds too quickly?
      5. ... not have a toxic metabolite?
3. Computational methods of attack
   1. Energy functions
   2. Probability
   3. Modeling and Simulation
   4. Artificial intelligence

1. Problem
   1. Resource constraints
      1. Time-consuming experiments with many manipulations
      2. Fixed personnel, space
      3. High cost of materials needed for experiment
   2. High variability
   3. Indirect measurements
2. One solution - surface-fitting methods of analysis (Other solutions are needed)
   1. Finding appropriate surface to fit
   2. How many and which data points to collect

Research Scientist
Agouron Pharmaceuticals, Inc

Hot Areas

Here are some hot areas of employment suitable for graduate and undergraduate mathematics, with the scientific areas needed (taken from that talk I gave last summer). Right now there is a big demand for statisticians who have the ability to program in SAS.

Bioinformatics - find patterns in chemical/biological information

• genetics
• biochemistry
• synthetic chemistry
• databases
• artificial intelligence
• computer programming (Oracle, C++)

• statistics, including design and analysis of (human) experiments
• probability (especially distributions)
• computer programming (SAS)
• basic biology and chemistry, information about well-known drugs

• biochemistry
• chemistry
• computer programming (C++ or C)
• artificial intelligence
• statistics
• probability
• numerical methods

    Return to profile of Laura Bloom