Notes:
idX tests for the following modifications, in addition to those indicated above:
- ammonia-loss at peptide's N-terminal glutamine;
- cystine at cysteine;
- deamidation at asparagine;
- dimethylation at argninine;
- hydroxyproline formation (collagens);
- hydroxylysine formation (collagens);
- oxidation at methionine;
- protein N-terminal acetylation;
This web site allows you to use idX for demonstration purposes. It is limited to analyzing the first 15,000 spectra
in a data file and the total number of simultaneous searches on the site is also limited. No spectrum files or output files are stored
on the site for any longer than necessary: for most sessions these files are deleted almost immediately. Some information is retained
by the web server's log for security purposes and a record of the start and finish times of each idX run is also maintained. No cookies
are used by this site.
idX is not an "open" search engine. It performs PSM assignments using peptide sequence kernels
that have a combination of curated and uncurated modification annotations. The 9 billion PSM asignments in GPMDB have been used
to guide this process. This sort of curation is technically a form of artificial intelligence.
idX
is open-source software. It is written in Python3 and it is available at
GitHub.
idX was designed to replace most of the arithmetic operations in conventional search engines with set-based mathematics.
Python's built-in
dictionary objects are used to speed up many of the set comparison operations. idX is CLI software:
this web site provides a simple, secure wrapper allowing you to use it remotely.
idX produces much less greenhouse gas than conventional search engines. Since it only requires one processor
and it performs the analysis quickly, it requires much less electricity to generate results.
idX will appear to run faster with large data sets than it does with small ones. An MGF file with 10,000 MS/MS spectra
will generate a report in about 90 seconds, while an MGF with only 1 MS/MS spectrum will take 40 seconds to generate a report.
This behavior is because a significant amount of information is loaded from disk before an analysis can be performed, resulting in
an overhead time of nearly 40 seconds for each run. Once that information is loaded, the idX algorithm can sort through
complicated lists of PTMs very quickly, resulting in relatively short total run times for large data sets.
