Re: QSAR - Rank order of activity data

Jim,

The simplest solution is to z-score the activity data [ (y-x)/s ]. This gives a continuous function so the, "binning issue is not a problem", but this is not going to alter the rank-ordering of the drugs from the order in the simple sort.

Perhaps some expansion concerning the information you hope to extract from the modified data that is not available from a sorted list would be of some help. Is your concern only that experimental variation may not indicate the same rank order for each measurement, or are there additional issues that you hope to resolve?

Are you attempting to derive a classification scheme for the activity in question or are you only interested in order of activity issues among the drugs in your dataset?

L. Mark Hall
Hall Associates Consulting

----- Original Message -----
  From: james.metz{=}abbott.com
  To: qsar_society=¬celrys.com
  Cc: james.metz[A]abbott.com
  Sent: Thursday, June 19, 2003 8:36 PM
  Subject: QSAR - Rank order of activity data

  QSAR Society Colleagues,

          I would like to convert biological activity or molecular property data to a rank order e.g., convert IC50 or p(IC50) data to a
  statistically meaningful rank order.

          This may sound like a problem with a simple answer, but apparently is not for several reasons:

          1) The trivial approach of simply sorting compounds by p(IC50) and then assigning a rank leads to artificial rank distinctions among
  compounds that given the inherent experimental variability of the data is probably not justified.

          2 If one had replicates for every IC50, one could attempt to do T-tests between every IC50, and assign a new rank if the statistic exceeds the 95% confidence level. However, I do not have replicates for every measured IC50 value.

          3) If one attempts to normalize the IC50 distribution and then assign ranks, perhaps, on the basis of mean +/- n*sigma values, this
  creates "somewhat" artificial ranks where some compounds may have been right at the border of the mean + sigma, or mean + 2*sigma, etc.

          If anyone has ideas, experience, or suggestions on this topic, I welcome your comments !

          Regards,
          Jim Metz

  James T. Metz, Ph.D.
  Research Investigator Chemist

  GPRD R46Y AP10-2
  Abbott Laboratories
  100 Abbott Park Road
  Abbott Park, IL 60064-6100
  U.S.A.

  Office (847) 936 - 0441
  FAX (847) 935 - 0548

  james.metz : abbott.com
Received on 2003-06-19 - 17:32 GMT