High-throughput Proteomics for Biomarker Discovery

The ability to reliably, robustly, and reproducibly detect quantitative protein abundance from clinical patient samples would have many applications in both research and clinical medicine. With its throughput and dynamic range, Fourier-transform ion-cyclotron resonance mass spectrometry (FTICR-MS) is one analytical approach that is showing promise in this area.
We aim to develop rigorous computational methods required to successfully analyze mass spectrometry data. In collaboration with the lab of Richard Smith (Pacific Northwest National Laboratory), we also aim to apply these approaches directly to patient samples. We have already seen promising results in the study of urine from kidney transplant patients facing acute rejection and the blood from burn and trauma patients post-injury.


A general platform for quantitative measurement of protein levels from human samples would remove a key hurdle to robust protein biomarker discovery. By enabling diagnoses both to be made earlier and to be more accurate, biomarkers have the potential to simultaneously reduce costs and improve outcomes in clinical medicine. Development of a rigorous statistical and informatics foundation for analysis of single-channel and dual-channel FTICR-MS data is necessary to harness the potential of this technology.


We have developed methods for normalization and statistical inference for both peptide-level and protein-level data. We have applied these methods to datasets from trauma, burn, and kidney transplant patients.


Amit Kaushal
Yuping Zhang
Wenzhong Xiao
Ronald W. Davis

Dept. of Statistics, Stanford University
Rob Tibshirani

Pacific Northwest National Laboratory
Weijun Qian
David Camp
Richard Smith

Department of Nephrology, Stanford University
Tara Sigdel
Minnie Sarwal

University of Texas Medical Branch
Celeste Finnerty
March Jeschke
David Herndon