Scientific Business Intelligence using Pipeline Pilot Anneliese Appleton Accelrys, Sydney y What is Scientific Business Intelligence? Biz Analyst Management Scientist Engineer Biz Analyst Management Business Intelligence Scientific Business Intelligence Limited Personas Simple Data Structures Highly Repeatable Tasks Simple Distribution Format A B Many Personas Complex Data Structures Structured and Unstructured Highly Granular Tasks Building Complex Routines Automating Processes Distributing Process Templates Collaborating Prototyping Applications Distributing Applications Well defined Single Data Model Multiple Data Models Fully Integrated Financial Data Customer Data Alpha Numeric Numeric Chemical Text Biological Corporate Data Image 2008 Accelrys, Inc. 2
Research Innovation Cycle Distribute & Report Results Present & Review Results Identify New Opportunity Gather Data Generate Report Prepare Data for Experiment Reformat Results Collate Results The Innovation Cycle Apply Model to Experiment Validate Model Build Model Import Data Innovation Productivity 2008 Accelrys, Inc. 3 What is Pipeline Pilot? 2008 Accelrys, Inc. 4
What is Pipeline Pilot? 2008 Accelrys, Inc. 5 Pipeline Pilot Enables practice reports and web 2008 Accelrys, Inc. 6
Pipeline Pilot Enables practice reports and web 2008 Accelrys, Inc. 7 Pipeline Pilot Enables practice reports and web 2008 Accelrys, Inc. 8
Pipeline Pilot Enables practice reports and web 2008 Accelrys, Inc. 9 Pipeline Pilot Enables practices reports and web 2008 Accelrys, Inc. 10
Broad Domain Capabilities of Pipeline Pilot Chemistry / Chemical Sciences Biological Sciences Scientific Reporting Data and Application Integration ti Image Analysis Statistics and Data Modeling Document Search and Analysis Analytical / Lab Data Analysis Many tasks require working across domain boundaries 2008 Accelrys, Inc. 11 Pipeline Pilot Software Demonstration 2008 Accelrys, Inc. 12
Environmental Chemistry & Toxicology with Pipeline Pilot 2008 Accelrys, Inc. 13 Crystal formation reporting with Pipeline Pilot Mouse over changes Display window ~85% of the model correctly scored with probability >. 80 ~12% of not sure images ~3% of false positive 0 false negatives Current crystallisation ti detection ti methods based on visual inspection Many customers have tried all other methods and found none to be reliable for their purposes Customers are adamant that such a system be free of false negatives, some mechanism must be allowed for review of the classification results to manually re-classify those droplets which were misclassified by the system 2008 Accelrys, Inc. 14
Polymer Property prediction with Pipeline Pilot 2008 Accelrys, Inc. 15 Clinical Image Management with Pipeline Pilot 2008 Accelrys, Inc. 16
Scientific Business Intelligence with Pipeline Pilot Process Modeling Automation Decision Support Collaborate and Deploy Users can build science process models using dragand-drop OO technology Easy To Bring Together Multiple Tasks Into A Single Solution User Friendly Advance Modeling And Informatics Find Meaning In The Data Portal Based Deployment Enhances Collaboration & Knowledge Re-use Pipeline Pilot Enterprise Server 2008 Accelrys, Inc. 17