how X!tandem E-value is calculated

[dhirendra]

hi,
can anyone provide me detailed explanation about how the X!tandem e-
value is calculated and how good is it for small dataset of say 100
sequences?

[Brian Searle]

Hi,

Here's a short talk I gave years ago for a user meeting on X!Tandem:
www.proteomesoftware.com/XTandem_edited.pdf

Pages 11-14 discuss how the e-value is calculated. What do you mean
exactly by data set/sequences? The e-value is not dependent on the
number of spectra in a data set as individual spectra are not scored
together (although PeptideProphet will have a harder time trying to
make sense of small data sets). The e-value can be database dependent
and it may not work well for a 100 sequence FASTA database if you're
employing high-mass accuracy peptide selection criteria.

Best, Brian

[Dhirendra Gupta]

thanks Brian,

       I saw the pdf but i am still unable to calculate evalue for some cases. Problem lies with small fasta database(100 sequences) where i dont get enough candidate peptides, in cases two or three. i am unable to draw a straight line of log(#results). I m surprised that x!tandem still provides evalues for such cases as i tested with the same small database. How does X!tandem tackle this problem?

2009/10/21 Brian Searle <sea...@gmail.com>

--
     Regards:

Dhirendra Kumar
Project assistant
IGIB

[Brian Searle]

I don't know what X!Tandem will do in that case. I suggest that you
contact Ron directly and let us know! Here's his contact info:
http://www.brc.ubc.ca/drupal5/node/37

Best, Brian

On Oct 22, 9:31 pm, Dhirendra Gupta <dhirendra8ku...@gmail.com> wrote:
> thanks Brian,
>
>        I saw the pdf but i am still unable to calculate evalue for some
> cases. Problem lies with small fasta database(100 sequences) where i dont
> get enough candidate peptides, in cases two or three. i am unable to draw a
> straight line of log(#results). I m surprised that x!tandem still provides
> evalues for such cases as i tested with the same small database. How does
> X!tandem tackle this problem?
>

> 2009/10/21 Brian Searle <sear...@gmail.com>