TY - JOUR
T1 - Ion trace detection algorithm to extract pure ion chromatograms to improve untargeted peak detection quality for liquid chromatography/time-of-flight mass spectrometry-based metabolomics data
AU - Wang, San Yuan
AU - Kuo, Ching Hua
AU - Tseng, Yufeng J.
PY - 2015/3/3
Y1 - 2015/3/3
N2 - Able to detect known and unknown metabolites, untargeted metabolomics has shown great potential in identifying novel biomarkers. However, elucidating all possible liquid chromatography/time-of-flight mass spectrometry (LC/TOF-MS) ion signals in a complex biological sample remains challenging since many ions are not the products of metabolites. Methods of reducing ions not related to metabolites or simply directly detecting metabolite related (pure) ions are important. In this work, we describe PITracer, a novel algorithm that accurately detects the pure ions of a LC/TOF-MS profile to extract pure ion chromatograms and detect chromatographic peaks. PITracer estimates the relative mass difference tolerance of ions and calibrates the mass over charge (m/z) values for peak detection algorithms with an additional option to further mass correction with respect to a user-specified metabolite. PITracer was evaluated using two data sets containing 373 human metabolite standards, including 5 saturated standards considered to be split peaks resultant from huge m/z fluctuation, and 12 urine samples spiked with 50 forensic drugs of varying concentrations. Analysis of these data sets show that PITracer correctly outperformed existing state-of-art algorithm and extracted the pure ion chromatograms of the 5 saturated standards without generating split peaks and detected the forensic drugs with high recall, precision, and F-score and small mass error.
AB - Able to detect known and unknown metabolites, untargeted metabolomics has shown great potential in identifying novel biomarkers. However, elucidating all possible liquid chromatography/time-of-flight mass spectrometry (LC/TOF-MS) ion signals in a complex biological sample remains challenging since many ions are not the products of metabolites. Methods of reducing ions not related to metabolites or simply directly detecting metabolite related (pure) ions are important. In this work, we describe PITracer, a novel algorithm that accurately detects the pure ions of a LC/TOF-MS profile to extract pure ion chromatograms and detect chromatographic peaks. PITracer estimates the relative mass difference tolerance of ions and calibrates the mass over charge (m/z) values for peak detection algorithms with an additional option to further mass correction with respect to a user-specified metabolite. PITracer was evaluated using two data sets containing 373 human metabolite standards, including 5 saturated standards considered to be split peaks resultant from huge m/z fluctuation, and 12 urine samples spiked with 50 forensic drugs of varying concentrations. Analysis of these data sets show that PITracer correctly outperformed existing state-of-art algorithm and extracted the pure ion chromatograms of the 5 saturated standards without generating split peaks and detected the forensic drugs with high recall, precision, and F-score and small mass error.
UR - http://www.scopus.com/inward/record.url?scp=84924041362&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84924041362&partnerID=8YFLogxK
U2 - 10.1021/ac504711d
DO - 10.1021/ac504711d
M3 - Article
C2 - 25622715
AN - SCOPUS:84924041362
SN - 0003-2700
VL - 87
SP - 3048
EP - 3055
JO - Analytical Chemistry
JF - Analytical Chemistry
IS - 5
ER -