what do workspaces do? can we run without them?

[hguturu]

What do workspaces actually do for tools like peptideprophet? Is there a way we can run without them (even if there is a performance hit)?

Looking at the issues like #61 and #107 it looks like you need to make a separate folder and workspace for each instance, which makes running the jobs a bit cumbersome.

I was also curious if there was a way to run peptideprophets outside of philosopher or tpp similar to percolator (where it takes a simple tsv/pin file) rather than a pepXML?

[prvst]

The reason we need one is that the workspace holds temporary data, metadata, and real results from the analysis. The program needs to keep track of the data while it filters, and alters each experiment results. Moreover, the program also collects vital information from your system to help with the processing, things like what kind of system you have, and where is the system temporary folder. This is also one of the reasons we need for the Prophets because they are not actually executed in the workspace folder, but in a separate location. At this moment, it's not possible to run the Prophets outside Philosopher without doing extensive alterations, you have to choose between using it in Philosopher, or with TPP. We actually run the Prophets in large batches, and in parallel, sometimes having dozens of instances running simultaneously, so the workspace should not cause you problems.

[hguturu]

This makes sense. I was just hoping for something a little less coupled to behind the scenes magic of workspaces to reduce the extra steps to run a big job.

Do you also know if there is a way to run peptideprophet with a simpler input than pepXML? I tried using the tsv generated by MSFragger using tsv_pepxml_pin, but that didn't produce the same output. I am also curious if we can re-run the final pepXML from running FragPipe or we need to grab the pepXML straight from MSFragger since the final pepXML of FragPipe has gone through additional annotations.

[prvst]

As far as I know, the maintainers of PeptideProphet decided that a pep.xml produced by a search engine program would be the only acceptable input, if any other format is accepted, I'm not aware of it.

[hguturu]

Makes sense. I figured as much, but was hoping there was a hidden feature that allowed that without dealing with xml.