Dec 8, 2016 Meeting Notes

From PHUSE Wiki
Revision as of 14:36, 8 December 2016 by William.houser (talk | contribs) (Meeting minutes of Nonclinical_Script_Assessment_Project)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search

Meeting minutes of Nonclinical_Script_Assessment_Project

Dec 8, 2016

Present First Name Last Name Affiliation
Yes Kevin Snyder FDA
Elaine Thompson FDA
Kevin Cahill EPL
Bill Varady Boehringer-Ingelheim
Bob Dorsam FDA
Yes Bob Friedman Xybion
Chris Eley Pfizer
Yes Stephen MacMannis Pfizer
Fred Mura PDS
Yes Anthony Fata PDS
Yes Hanming Tu Accenture
Yes Bill Houser Bristol-Myers Squibb
Carol Gleason Bristol-Myers Squibb
Yes Atul Mishra Pfizer
Peter Scaefer Validated Cloud Applications

Action Items from previous meetings:

  • We should reach out to Pinnacle 21 to join us - Bill sent email Aug 17, 2016 and Sept 22 asking if they would be interested in in joining us. - Tony will ask Mike to see if he is communicating with Max at Pinnacle 21. - no response.
    • Their experience would help us
    • They could benefit from our contributions
  • Bill Houser will create list of principles that we agree to as general practice.- not done yet.
    • Kevin Snyder found that to analyze the PDS data we need to look at PCTPTNUM to get the hour after dose, but this doesn't work with the Instem dataset. For the Instem script we need PCELTM and PCRFTDTC. He suggested it would probably be worth enabling the script to work with either dataset.
    • We discussed if we should preserve the SEND values or modify them as we prepare the information for analysis. We concluded that we should preserve the SEND values and copy or transform them to variables we use for analysis.
    • As we proceed, we agreed to try to create generalized functions in a library. (concept similar to Unix commands in that they are simple and can be combined in powerful ways).
    • New Function Idea: create a function that will pull rows from a .xpt file that match a subjects defined by a characteristic in another .xpt file (e.g., get the body weight data from only male animals)
  • All should try to the TK analysis script with our private data to see what we can learn.
  • We should each try the script on our own data and share the results
    • Stephen tried and encountered an error reading the xpt file. There are multiple analytes in his data, but the data frame only had one analyte.
    • Kevin has successfully used it on a study with multiple analytes.
    • Stephen will try some more.
  • Kevin showed the Krona plotting tool for browsing hierarchical data: https://github.com/marbl/Krona/wiki
  • Hanming volunteered to integrate a parameter file (using .yml format) into one of the scripts.

Meeting Notes:

  • Bill contributed some SAS scripts that he previously wrote
    • SEND3.0-to-print 0.5.sas for creating a PDF file of the contents of SEND data tabulations
    • SEND3.0-to-xml-v0.5.sas for creating an xml representation of SEND datasets
    • SEND3_0-from-xls-v0_12.sas for creating *.xpt from Excel files
  • Hanming walked us through an R script that loads data into Oracle.
  • Bob Friedman talked through his contributed/scriptathon2014/scripts/targettargeto8.R
  • Hanming found an article describing the differences between between "=", "->", and "->>": https://stat.ethz.ch/R-manual/R-devel/library/base/html/assignOps.html
  • Bill asked if anyone had an example that would be useful to take data from a pdf table of Toxicokinetic data to create SEND PC and PP.
    • Kevin said that text manipulation is not a strength of R and that python would be better.
    • Kevin has found tabula (http://tabula.technology/) to be useful in converting PDF tables to csv for subsequent analysis.

Action Items:

  • Peter will consider the possibility of generating test data sets for toxicokinetics data in accordance with the SEND standard with the goal of identifying all of the different ways that this data could be "correctly" encoded and determining whether a universal analysis routine could be performed successfully across all possible interpretations of the standard.
  • Bill will ask Gitte if the "Application of SEND Data for Analysis" has an report that would be useful for the scripting group to create from SEND.
  • Hanming will contribute the script he wrote to load data into Oracle including a reference to the article that guided him.