You should be able to upload a Mascot .dat file, if you navigate to it using the data pipeline "Process and Upload Data" button.

On the radical discrepency between v1.7 and v2.1:
1. How many peptides do you have in the v1.7 data with PeptideProphet probability greater than 0.05?
2. If you look at the PeptideProphet details graphs (button on the peptides page), how do the models look for v1.7 v. v2.1? (Assume at least on valid model for v1.7 and none for v2.1).
3. Do the X! Tandem "Expect" scores look much different between v1.7 and v2.0? Sort by this score, and see how many you have less than 0.01 (smaller is better).
4. If you replace the v2.1 PeptideProphet executable with the v1.7 version, how do the results change?

If you feel comfortable attaching the .xtan.xml and .pep.xml files from your results, please do.

--Brendan

The answers to the questions are below, plus I am attaching the .xtan.xml and .pep.xml files for the v1.7. Since there's only option for 3 files, I'll attach the v2.1 files to another response.

On the radical discrepency between v1.7 and v2.1:
1. How many peptides do you have in the v1.7 data with PeptideProphet probability greater than 0.05?

Peptides with PP probability > 0.05: 1733
Peptides with PP probability < 0.05: 510

The highest probability is 0.9988, and there are 194 with probabilities > 0.9


2. If you look at the PeptideProphet details graphs (button on the peptides page), how do the models look for v1.7 v. v2.1? (Assume at least on valid model for v1.7 and none for v2.1).
Umm, not so good. I did screen shots of the detail graphs and could send jpgs, but I'll attach the .xtan.xml and .pep.xmls. The table for v1.7 model is:

v1.7
Minimum     Sensitivity    Error rate
probability
0.00    1.0000    0.6862
0.05    0.9917    0.5972
0.10    0.9750    0.5631
0.15    0.9481    0.5307
0.20    0.9004    0.4851
0.25    0.8361    0.4275
0.30    0.7718    0.3697
0.35    0.7255    0.3289
0.40    0.6916    0.3015
0.45    0.6636    0.2813
0.50    0.6176    0.2530
0.55    0.5773    0.2303
0.60    0.5257    0.2041
0.65    0.4794    0.1829
0.70    0.4161    0.1558
0.75    0.3561    0.1325
0.80    0.2825    0.1043
0.85    0.2123    0.0776
0.90    0.1402    0.0508
0.95    0.0704    0.0277
0.98    0.0226    0.0073
0.99    0.0170    0.0040


3. Do the X! Tandem "Expect" scores look much different between v1.7 and v2.0? Sort by this score, and see how many you have less than 0.01 (smaller is better).

The lowest E is 0.011.
I compared the E values for the scans that came up in v2.1 and are in the spreadsheet E values compared. They were in pretty good agreement.

4. If you replace the v2.1 PeptideProphet executable with the v1.7 version, how do the results change?
How would I do this?

Thanks for the help!

Mary Davis

The files for v2.1 results are attached.

Mary,

I'm attaching the versions of PeptideProphet that we shipped with version 1.7 and with version 2.0. Assuming that you're running on Windows with a standard installation, do the following to try them. Depending on what version you first installed, your <labkey_root> might be set to c:\labkey or c:\program files\labkey server.

1. Copy <labkey_root>/bin/PeptideProphetParser.exe to <labkey_root>/bin/PeptideProphetParser21.exe to back up the current version.
2. Copy the attached PeptideProphetParser17.exe to <labkey_root>/bin/PeptideProphetParser.exe.

Then, try doing another search and comparing the results. You can also try the attached PeptideProphetParser20.exe and compare its results.

Thanks,
Josh

Josh,

Thanks for the files. I ran the analysis with PeptideProphet1.7 and the results look more like the runs done in March and April with CPAS v1.7. There were 1027 peptides with PP probability greater than 0.05, and an additional 33 with probabilities of -2.000 (they had smallish probabilities in the earlier analysis).

I also tried running with the previous tandem.exe and that made a little difference, only 774 peptides with probability greater than 0.05.

I looked at the logs and the xinteract build also changed between CPAS1.7 and 2.1, and the options selected. Would this have made the difference? I looked at the documentation on the TPP site and couldn't figure out what the -Opt setting does (I see that -Op says to run ProteinProphet after, but what is the t or is this different)?

I am attaching the logs and on another reply will attach the output from using the PeptideProphet exe from CPAS v1.7. The file CPAS2_1_PepProph_1_7_log.txt is the one using PeptideProphet exe from CPAS v1.7 in current CPAS.

Thanks

Mary

This is the output from running the PeptideProphet exe from v 1.7 in CPAS 2.1. (Because these are MHC peptides, I am not looking at protein level analysis.)

Thanks again

Mary