Presentation is loading. Please wait.

Presentation is loading. Please wait.

Using SAS Stored Processes and the SAS Portal for Delivering Statistics to Drug Discovery Volker Harm PhUSE/PSI One-day Event 2009, Marlow.

Similar presentations


Presentation on theme: "Using SAS Stored Processes and the SAS Portal for Delivering Statistics to Drug Discovery Volker Harm PhUSE/PSI One-day Event 2009, Marlow."— Presentation transcript:

1 Using SAS Stored Processes and the SAS Portal for Delivering Statistics to Drug Discovery Volker Harm PhUSE/PSI One-day Event 2009, Marlow

2 Delivering Statistics to Drug Discovery PhUSE/PSI One-Day Event 2009, Marlow Page # 2 Using SAS Stored Processes and the SAS Portal for Delivering Statistics to Drug Discovery

3 Delivering Statistics to Drug Discovery PhUSE/PSI One-Day Event 2009, Marlow Page # 3 Introduction Drug Discovery Statistics -No regulated environment that requires analyses done by a statistician -Small dislocated working groups with very special requirements -Application development in large scale economically unfeasible Supporting Drug Discovery -Build up a reputation on applying statistics to research problems -Give statistical advice -Write statistical report -Develop SAS program, evaluate, discuss with scientists -Establish a standard procedure of evaluating experiments

4 Delivering Statistics to Drug Discovery PhUSE/PSI One-Day Event 2009, Marlow Page # 4 Introduction (2) The Statistics Portal -SAS®9: chance to a a step further -SAS® Enterprise Guide, SAS® Stored Processes, the SAS® Stored Process Application and the SAS Information Delivery Portal -prototype application, which can show the feasibility of a Statistics Portal -Given the wealth of standard routines at hand, which can easily be adapted, centrally deployed and managed, this truly could be a means to deliver statistics to the scientists.

5 Delivering Statistics to Drug Discovery PhUSE/PSI One-Day Event 2009, Marlow Page # 5 SAS® BI Infrastructure Used As straightforward as possible Components as provided by the standard installation SAS Enterprise Guide to create the stored processes. Basic functionality of the SAS Information Delivery Portal -Links -Collections -Ability to execute stored processes

6 Delivering Statistics to Drug Discovery PhUSE/PSI One-Day Event 2009, Marlow Page # 6 Statistical Support - The Traditional Way

7 Delivering Statistics to Drug Discovery PhUSE/PSI One-Day Event 2009, Marlow Page # 7 The Statistical Report The problem was analyzed and a report was written. The report states in its introduction: -“As part of cytotoxicology experiments IC50 values of different compounds are determined. In addition to different compounds, the IC50 value of Menadione (positive control) is determined per experiment. -Current practice is that the validity of the experiment is assessed by comparing the IC50 value of Menadione to the mean ±2std of historical values of IC50 values of Menadione. Values that are outside the boundary indicate a possible validity issue. -This report introduces control charts for this process, and will recommend some small changes to the current practice. Furthermore, control charts are given for selected parameters.”

8 Delivering Statistics to Drug Discovery PhUSE/PSI One-Day Event 2009, Marlow Page # 8 The Statistical Report (2) And the solution is described as: -“We will create a control chart for individual measurements and moving ranges. These are special forms of so-called “Shewhart charts”, that are used in process control. The respective IC50 values are log-transformed to approach a normal distribution of the measurements. -Two graphs are provided. One figure contains the individual measurement chart (I-chart), in which the individual (log- transformed) IC50 values and the control ranges (control limits) are displayed. The moving range chart (R-chart) contains the ranges of the two most recent measurements. Also in this plot control ranges are displayed.”

9 Delivering Statistics to Drug Discovery PhUSE/PSI One-Day Event 2009, Marlow Page # 9 The SAS Program

10 Delivering Statistics to Drug Discovery PhUSE/PSI One-Day Event 2009, Marlow Page # 10 Control Charts for CytotoxicologyExperiment

11 Delivering Statistics to Drug Discovery PhUSE/PSI One-Day Event 2009, Marlow Page # 11 The Standard Procedure As this is easily done as a one time task, in a routine process it is a bit more tedious -There are four cell lines, for which these values are determined. -Each time an experiment is conducted, new values will be added and new control ranges should be determined. In the traditional way that would mean for the scientist: -Deliver the new data to the Global Drug Discovery Statistics Department -Wait for the evaluation

12 Delivering Statistics to Drug Discovery PhUSE/PSI One-Day Event 2009, Marlow Page # 12 The Standard Procedure (2) And for us: -Add the new data -Run the program -Deliver the results to the scientists.

13 Delivering Statistics to Drug Discovery PhUSE/PSI One-Day Event 2009, Marlow Page # 13 Extending Statistical Support – The SAS Portal Solution

14 Delivering Statistics to Drug Discovery PhUSE/PSI One-Day Event 2009, Marlow Page # 14 SAS Stored Process Is a SAS program hosted on server described by metadata -Name -Description -Keywords -Input Parameters -Output Options -Execution Server -Security Options

15 Delivering Statistics to Drug Discovery PhUSE/PSI One-Day Event 2009, Marlow Page # 15 SAS Stored Process “New IC50”

16 Delivering Statistics to Drug Discovery PhUSE/PSI One-Day Event 2009, Marlow Page # 16 Requirements While in the traditional way the main focus lies on the number crunching, in providing an application used by the scientist, we also have to consider the requirements on the data from the scientists view. Therefore the first stored process we have to develop is one to get new data not the highly sophisticated evaluation. This needs a bit of requirements analysis together with the user.

17 Delivering Statistics to Drug Discovery PhUSE/PSI One-Day Event 2009, Marlow Page # 17 Requirements (2) Whereas in our one-off program only the statistically relevant variables IC50 for the IC50 value and nr for the number of the value appeared Our stored process for adding values to the database handles all the parameters relevant to the scientist.

18 Delivering Statistics to Drug Discovery PhUSE/PSI One-Day Event 2009, Marlow Page # 18 The Generated Code for the Stored Process Stored Process Parameter Dictionary Generated Global Macro Variables Our SAS Code -Define the Stored Process Name -Do the Includes -Get the Data from the Application -Update the IC50 database -Report the Action -Call the Macros The Generated Stored Process End

19 Delivering Statistics to Drug Discovery PhUSE/PSI One-Day Event 2009, Marlow Page # 19 Stored Process Parameter Dictionary

20 Delivering Statistics to Drug Discovery PhUSE/PSI One-Day Event 2009, Marlow Page # 20 Generated Global Macro Variables

21 Delivering Statistics to Drug Discovery PhUSE/PSI One-Day Event 2009, Marlow Page # 21 Get the Data from the Application

22 Delivering Statistics to Drug Discovery PhUSE/PSI One-Day Event 2009, Marlow Page # 22 Update Database and Report Action

23 Delivering Statistics to Drug Discovery PhUSE/PSI One-Day Event 2009, Marlow Page # 23 Features of the Stored Process Required fields are marked with an asterisk. For Parameter CellLine a list of valid values is used. Dates can be entered by using a calendar applet. Standard SAS execution options can be selected. The added entry is reported in a separate output window.

24 Delivering Statistics to Drug Discovery PhUSE/PSI One-Day Event 2009, Marlow Page # 24 Stored Process „New IC50“ The item of the record to add are presented Required items are marked with an asterisk

25 Delivering Statistics to Drug Discovery PhUSE/PSI One-Day Event 2009, Marlow Page # 25 Stored Process „New IC50“ (2) List of values are for input fields „Cell Line“ and „Valid?“

26 Delivering Statistics to Drug Discovery PhUSE/PSI One-Day Event 2009, Marlow Page # 26 Stored Process „New IC50“ (3) Dates can be entered using a calendar applet

27 Delivering Statistics to Drug Discovery PhUSE/PSI One-Day Event 2009, Marlow Page # 27 Stored Process „New IC50“ (4) Standard SAS execution options can be selected

28 Delivering Statistics to Drug Discovery PhUSE/PSI One-Day Event 2009, Marlow Page # 28 Stored Process „New IC50“ (5) The added item is reported in the output window.

29 Delivering Statistics to Drug Discovery PhUSE/PSI One-Day Event 2009, Marlow Page # 29 Stored Processes „Display Cell Line“ and „Display Control Charts“ Presentation of several reports -Graph and data for one cell line -Control charts for all cell lines

30 Delivering Statistics to Drug Discovery PhUSE/PSI One-Day Event 2009, Marlow Page # 30 “Display Cell Line” selects data of one cell line from the database calculates the control values from the last 50 valid entries and displays the control charts as well as the data of the cell line.

31 Delivering Statistics to Drug Discovery PhUSE/PSI One-Day Event 2009, Marlow Page # 31 Stored Process „Delete IC50“ Delete a single record by specifying -the cell line and -the experiment date This stored process is used in the rare case, when the scientist detects an error in the entries

32 Delivering Statistics to Drug Discovery PhUSE/PSI One-Day Event 2009, Marlow Page # 32 The SAS Portal Collects and presents the four stored processes Makes a complete application Accessible from everywhere in the enterprise Centrally maintained and deployed

33 Delivering Statistics to Drug Discovery PhUSE/PSI One-Day Event 2009, Marlow Page # 33 SAS Stored Process Web Application

34 Delivering Statistics to Drug Discovery PhUSE/PSI One-Day Event 2009, Marlow Page # 34 Conclusion I think stored processes really a new dimension to statistical support. Of course it is more a matter of presentation than of statistics. But if we want the scientist to “Do more statistics!”, that is what we have to do.

35 Delivering Statistics to Drug Discovery PhUSE/PSI One-Day Event 2009, Marlow Page # 35 Recommended Reading Cooman, Franky De and Veerbeck, Rudy. 2006. “Strategic roadmap for the IT support of the analysis and interpretation of data in drug discovery”, Paper TS08, PhUSE 2006 Hall, Angela and Miles, Brian. 2007. “SAS® Information Delivery Portal - A Tutorial”, Paper 214-2007, SAS Global Forum 2007 Pope, David (2007). “Empowering Your SAS Business Intelligence End Users via a SAS Stored Process”, Paper 038-2007, SAS Global Forum 2007

36 Delivering Statistics to Drug Discovery PhUSE/PSI One-Day Event 2009, Marlow Page # 36 Recommended Reading (2) Pratter, Frederick E. (2007). “Delivering Dynamic Content with the SAS® Stored Process Web Application”, Paper 231-31, SUGI 31 Pratter, Frederick E. (2007). “SAS Enterprise Guide and Stored Processes”, Paper 218-2007, SAS Global Forum 2007 SAS Institute Inc. 2007. SAS Institute white paper. “SAS® Stored Processes - Introduction and Overview”

37 Thank you for your attention

38


Download ppt "Using SAS Stored Processes and the SAS Portal for Delivering Statistics to Drug Discovery Volker Harm PhUSE/PSI One-day Event 2009, Marlow."

Similar presentations


Ads by Google