openmolecules.org Forum: Functionality » How to calculate selectivity score?

Home » DataWarrior » Functionality » How to calculate selectivity score? (How to calculate selectivity score from kinase residual activity data?)

Show: Today's Messages :: Polls :: Message Navigator

How to calculate selectivity score? [message #212]

Fri, 30 December 2016 06:08

jeetu270
Messages: 22
Registered: October 2015
Location: INDIA

Junior Member

Hi Thomas,

There is an option for calculating selectivity score in DataWarrior, which I think is Gini coefficient (if I am not wrong). I tried calculating this selectivity score for an inhibitor by giving the % residual kinase activities as the input column. But after running this calculation, its showing NaN in all the rows. Can you please guide me how to calculate this score since there is no help on this topic in your website.

Thanks for the great software..

Jitender

Report message to a moderator

Re: How to calculate selectivity score? [message #214 is a reply to message #212]

Sat, 31 December 2016 17:24

thomas
Messages: 729
Registered: June 2014

Senior Member

Hi Jitender,

I tried with some values and even empty cells, but always got numerical scores. There seems to be an issue, however, with logarithmical columns. The Gini logic should not be applied to logarithms, because the concept is 0-based. Currently, if a column is defined to be logarithmically treated, then then the Gini calculation uses the logarithms, which it shouldn't. I will solve that!

To solve your problem you may send me a data file and tell me which columns you use to generate the Gini score.

Regards, Thomas

Report message to a moderator

Re: How to calculate selectivity score? [message #215 is a reply to message #214]

Mon, 02 January 2017 05:02

jeetu270
Messages: 22
Registered: October 2015
Location: INDIA

Junior Member

Thanks for the reply Thomas.

Actually as far as I understand the calculation of Gini score in the article, we need to input only a column containing the % residual activities for individual kinases. I tried with the example given in that paper (Table 1), and in their excel sheet only. It worked and calculated a single Gini score value. But the same % residual activities column (Column B) when I gave as input to DataWarrior, it showed Nan in all rows.

I am attaching the required files here, Please check..

Thanks
Jitender

Attachment: Gini Coefficient.pdf
(Size: 189.68KB, Downloaded 1164 times)
Attachment: jm070562u_si_004.xls
(Size: 340.50KB, Downloaded 370 times)
Attachment: Residual Activities.xlsx
(Size: 8.56KB, Downloaded 341 times)

Report message to a moderator

Re: How to calculate selectivity score? [message #217 is a reply to message #215]

Mon, 02 January 2017 05:31

jeetu270
Messages: 22
Registered: October 2015
Location: INDIA

Junior Member

Hi Thomas,

Also I tried giving the % residual kinase activities for an inhibitor in a single row as shown in the attached excel file. It calculated and gave a Gini score value (0.25181) for the inhibitor but it is not matching with the one I calculated from the automated excel sheet provided by the author of the article (0.578)?

I think I am not providing the input in the right manner. Please help.

Thanks
Jitender

Attachment: Gini_Coefficient_Calculation.dwar
(Size: 2.34KB, Downloaded 754 times)
Attachment: Residual Activities_Transposed.xlsx
(Size: 8.66KB, Downloaded 722 times)

Report message to a moderator

Re: How to calculate selectivity score? [message #223 is a reply to message #217]

Sun, 05 February 2017 22:36

thomas
Messages: 729
Registered: June 2014

Senior Member

Hi Jitender,

sorry for the late reply. I checked the paper, your files and the current Gini calculation method in DataWarrior. The main reason for the discrepancy is that the sample file from the paper contains effect values. Their published method converts them into inhibitions by basically inverting the values (inhibition=100-effect). Then they limit the value range by replacing any value beyond the allowed range (0-100) to the respective range limit (100 or 0). Into the bargain my method of calculating the areas was a little different. I changed my code to use their so-called trapez method. I also have a new option to invert the data (100-x). However, I will not do the correction to move values into the allowed range, because that would remove part of the noise in an unsymmetrical fashion. The update is due soon.

Thomas

Report message to a moderator

Re: How to calculate selectivity score? [message #226 is a reply to message #223]

Mon, 06 February 2017 09:21

jeetu270
Messages: 22
Registered: October 2015
Location: INDIA

Junior Member

Thanks for the explanation Thomas...

Jitender

Report message to a moderator

Previous Topic:	Using Set Value Range option
Next Topic:	Option to hyperlink data.

Goto Forum:

-=] Back to Top [=-

[ Syndicate this forum (XML) ] [

] [

]

Current Time: Mon Apr 07 11:11:26 CEST 2025

Total time taken to generate the page: 0.03665 seconds