header area ProtoLifeProtoLife
About Us*Technology*Contact

 
 
Overview

Predictive Design Technology™ (PDT) makes it possible to find optimal targets in complex experimental spaces that are too large or expensive to search exhaustively.


PDT is unique:
  • While many experimental design methods are one-off methods for telling you which experiment to do next, PDT is embedded in an automated, iterated high-throughput experimentation cycle. It closes the loop by continuously designing round after round of experiments until the screening target is reached.
  • PDT is a proprietary meta-modeling technology, choosing from a variety of predictive modeling techniques that efficiently find optimal targets in complex search spaces.
  • Proprietary algorithms make intelligent decisions at each experimental iteration: PDT chooses the optimal type and complexity of models to use for each specific experimental system, and makes smart tradeoffs between exploration and exploitation, exploring new areas of the experimental space while exploiting accumulating data from successive experimental runs.
  • PDT is optimized for spaces with many different parameters and multilevel qualitative variables.

PDT goes beyond...
  • ... traditional design of experiments (DoE) (for example, JMP® or Design-Expert® software): Traditional DoE methods work when the experimental space is small enough to search exhaustively, or when interactions among system components are weak enough that the space can be simplified. PDT works when these conditions fail, that is, when components have strong, complex interactions.
  • ... genetic algorithms (GAs): PDT is a sophisticated evolutionary algorithm with much more power than traditional GAs (see PDT in Action.)
  • ... neural networks: These are just one component of PDT's modeling toolkit; PDT intelligently chooses among neural networks and other (both well-known and proprietary) modeling techniques to find optimal targets with unprecedented efficiency.
  • ... combinatorial chemistry: PDT does not assay complete libraries; it is a tool for intelligently sub-sampling experimental spaces that are too large and complex to examine exhaustively.
  • ... chemoinformatics or quantitative structure-activity relationship (QSAR) models: PDT exploits any available structural information, but can find optimal targets even when such information is unavailable. Chemoinformatics and QSAR rely on theoretical information about molecular compounds; when the system being searched is complex enough, these methods are often unreliable. PDT's predictive models are built directly on the behavior of each experimental system they optimize.
  • ... mining of genomic, proteomic, X-omic data: These methods are not typically iterative or adaptive.
Microarray photo
 
 
Copyright © 2010, ProtoLife, Inc. About Us   |   Technology   |   Contact   |   Home Website: www.protolife.com  /  e-mail: info@protolife.com