Sequential Screening Process

Introduction

To increase the value of screening efforts, there is a growing movement towards in silico techniques as a means to select and prioritize which molecules to screen next. Here, we discuss the combined in vitro-in silico sequential screening process. By screening a relatively small training set of compounds and building a model that differentiates between high-activity and low-activity molecules, any compound collection, commercially available or virtual, can be screened in silico to prioritize the next iteration of screening and model building. This approach can accelerate the rate of drug discovery, and save costs by directing in vitro screening efforts to the compounds with highest probability of success.

There are a number of problems sequential screening addresses:

Problem

Despite orders of magnitude expansion in high-throughput screening (HTS) over the past decade, fewer new chemical entities are making it to market.

Organizations have little to show for all of the innovations of the 1990s, at least in terms of new chemical entities. HTS, in particular, has not lived up to the expectations that greeted its initial adoption in the first half of the decade. As a field, HTS has matured, but lead attrition rates remain high, and little time has been shaved off R&D pipelines. There are several reasons for this:

  • The search space of compounds is estimated at more than 10^60. The ability to screen a million compounds instead of a thousand hardly begins to thoroughly map out this search space.
  • High throughput screening has been focused mostly on optimizing for potency and this is to the exclusion of other properties, such as ADME and Toxicity, which ultimately determine whether a compound will be viable as a drug. This leads to higher attrition rates downstream, with at least 4 out of 5 compounds failing in clinical trials.
  • In general, due to market demands and FDA regulations, for a new drug to make it to market, it must be safer and/or more effective than existing drugs within a given therapeutic class. Within a given class, the bar will continually be raised with each new product, making the probability of success continually drop over time. The first drug to market in a class is best with probability 1. However, as time goes by, the probability a given drug is better than all previous ones drops inversely with time. That is, given other attempts, the probability a new drug is better than all previous attempts is 1/n.
  • The cost of discovery has skyrocketed, driving discovery efforts more towards blockbuster drugs where the probability of success is low, as explained above.

Sometimes even enormous screens still fail to deliver viable hits.

Combinatorial chemistry can now synthesize a near-infinite number of compounds and the problem is narrowing down the list to a reasonable subset to build and screen.

Solution

Use sequential screening to more rapidly find better compounds with lower attrition rates and with lower costs.

  • The sequential screening paradigm can be thought of as a gradient search through the vast chemically-accessible search space.
  • Use sequential screening to simultaneously optimize for potency, ADME, and non-toxicity, delivering better compounds to the medicinal chemists for final optimization. With lower failure rates downstream, the cost of drug discovery will come down. Further, by using staggered sequential screening, it is possible to concurrently perform more assays with existing screening capacity, diversifying risk over more programs.
  • By lowering the attrition rate of compounds, there is a greater chance of delivering a drug that is better than existing ones.
  • Furthermore, as the cost of drug discovery comes down, smaller therapeutic classes can be addressed, opening up new markets that have few or no competing drugs to beat.

With ChemTree sequential screening it is possible to resurrect failed screens. If the assay is valid, and we have a distribution of activities among the inactives, then we can build a ChemTree predictive model to select a new round of compounds to screen with the high potency tails of the distribution greatly enhanced. These compounds can be selected from in house collections, third party vendors, or built through selective combinatorial synthesis. A more detailed example can be found here.

By combining the predictive power of sequential screening and the diversity of molecules available with combinatorial chemistry, it becomes possible to search through chemically accessible space in a focused manner -- iteratively building molecules that not only bind to the target, but also have positive drug-like properties.

Details

Sequential screening paradigm. diagram here.

With ChemTree multivariate analysis, it is possible to simultaneously optimize for multiple chemical endpoints within the same model. This is particularly useful in resolving issues of drugs failure downstream due to poor absorption, distribution, metabolism, elimination (ADME) or toxicity properties - where we employ sequential screening for predictive ADME/Tox. The ability to perform smaller, smarter screens can also bring massive savings when the cost of screening a compound is high.