% IMPORTANT: The following is UTF-8 encoded. This means that in the presence
% of non-ASCII characters, it will not work with BibTeX 0.99 or older.
% Instead, you should use an up-to-date BibTeX implementation like “bibtex8” or
% “biber”.
@ARTICLE{Holl:280903,
author = {Holl, Sonja and Mohammed, Yassene and Zimmermann, Olav and
Palmblad, Magnus},
title = {{S}cientific {W}orkflow {O}ptimization for {I}mproved
{P}eptide and {P}rotein {I}dentification},
journal = {BMC bioinformatics},
volume = {16},
issn = {1471-2105},
address = {London},
publisher = {BioMed Central},
reportid = {FZJ-2016-00614},
pages = {284},
year = {2015},
abstract = {Background: Peptide-spectrum matching is a common step in
most data processing workflows for massspectrometry-based
proteomics. Many algorithms and software packages, both free
and commercial, have beendeveloped to address this task.
However, these algorithms typically require the user to
select instrument- andsample-dependent parameters, such as
mass measurement error tolerances and number of missed
enzymaticcleavages. In order to select the best algorithm
and parameter set for a particular dataset, in-depth
knowledgeabout the data as well as the algorithms themselves
is needed. Most researchers therefore tend to use
defaultparameters, which are not necessarily
optimal.Results: We have applied a new optimization
framework for the Taverna scientific workflow management
$system(http://ms-utils.org/Taverna_Optimization.pdf)$ to
find the best combination of parameters for a given
scientificworkflow to perform peptide-spectrum matching. The
optimizations themselves are non-trivial, as demonstrated
byseveral phenomena that can be observed when allowing for
larger mass measurement errors in sequence databasesearches.
On-the-fly parameter optimization embedded in scientific
workflow management systems enables expertsand non-experts
alike to extract the maximum amount of information from the
data. The same workflows could beused for exploring the
parameter space and compare algorithms, not only for
peptide-spectrum matching, but alsofor other tasks, such as
retention time prediction.Conclusion: Using the optimization
framework, we were able to learn about how the data was
acquired as well asthe explored algorithms. We observed a
phenomenon identifying many ammonia-loss b-ion spectra as
peptideswith N-terminal pyroglutamate and a large precursor
mass measurement error. These insights could only be
gainedwith the extension of the common range for the mass
measurement error tolerance parameters explored by
theoptimization framework.},
cin = {JSC},
ddc = {004},
cid = {I:(DE-Juel1)JSC-20090406},
pnm = {511 - Computational Science and Mathematical Methods
(POF3-511) / 512 - Data-Intensive Science and Federated
Computing (POF3-512)},
pid = {G:(DE-HGF)POF3-511 / G:(DE-HGF)POF3-512},
typ = {PUB:(DE-HGF)16},
UT = {WOS:000360426000008},
pubmed = {pmid:26335531},
doi = {10.1186/s12859-015-0714-x},
url = {https://juser.fz-juelich.de/record/280903},
}