The Cytoprint image fingerprints and the image change signatures are used as database indexes. Atto’s Cytoprint pattern recognition technology is being developed into a database tool.
Automated Assay Image Analysis
Scientists can rapidly develop algorithms for assay image analysis. By training on example images from assays of positive and negative controls, the Cytoprint pattern recognition software can automatically identify and quantify the assay response. Using the trained assay response, the biological effect of unknown compounds or assay treatments can be rapidly assessed.
Training Cytoprints for assay image analysis can often result in more sensitive and more selective assays when compared to traditional image analysis. The shotgun feature extraction approach first applies a wide range of unbiased image measurements to each cell. Statistical analysis considers the information content in each measurement. The resulting combination of measurements that is used to score the assay will be performed with the optimum feature set, that may include morphological changes that are not included in a traditional image analysis algorithm that was developed using biological expectations.
Automated Cytoprint assay image analysis has been performed on a wide range of targeted cellular morphologies on data from a range of instruments.
Assay Optimization
Assay optimization can be as simple as trading off increases in assay selectivity and specificity for assay speed.
Assay optimization experiments such as ANOVA or factorial analysis can also be facilitated by using the Cytoprint technology to automatically create an unbiased assessment of the performance of each set of assay conditions. Automatically generating the best image analysis using the Cytoprint technology avoids the time, cost, complexity and distortion in the results that would come from having to manually tailor the image analysis to each different assay conditions.
Clustering and Data Mining Tools for Data Driven Assay Discovery
Scientists can use the Cytoprints and signatures to mine a database of assay results in order to discover novel similarities and differences in biological responses and cell behavior. For example, clustering Cytoprints can associate compounds that have a similar toxicity even if these compounds have widely different primary biological activities.
Multiplexed, Multiendpoint Assays
Atto is using data driven assay discovery to create multiplexed, multiendpoint assays. Different groups of positive control compounds that are run in the same assays can be used to create separate Cytoprint signatures for separate biological activities. Scoring an unknown result with each trained Cytoprint yields an assessment of the separate biological activity of that positive control group. Panels of several dozen related and unrelated endpoints are possible.
Copyright © 2004 Atto Bioscience All rights reserved