In the CPTAC consortium, we characterize tumors with both proteomic and genomic data. We use multi-omics data to reveal new hints at cancer therapies, like using DNA copy number alterations and their trans-effected proteins to create a signature of survival in ovarian cancer( Cell 2016). We are also interested in the function of post-translational modifications and how these intersect with DNA mutations. Finally, we are dedicated to open science and data sharing technologies.
- What mutations and proteome state are correlated with a good survival in uterine and endometrial cancer?
- How do mutations affect the behavior of post-translational modifications?
- Are protein complexes dysregulated in a consistent manner across tumor types?
Machine Learning and Forensics
Bioinformatics can be used to identify a victim, criminal, or unknown sample. A wide variety prosecutorial needs are not met by DNA evidence, and could be solved by proteomic evidence. Here, there exists less of a data and statistical framework for how one measures accuracy and validity, i.e. what would be admissible in a court of law ( JPR 2018).
- How can we measure and characterize human proteome variability?
- What is a statistically confident identification in a legal situation?
- Can we distinguish different human tissues at crime scenes?
Mass Spectrometry Identification
I have a sustained interest in fundamental algorithms for the identification of mass spectra. - phosphorylated peptides ( JPR 2008)- proteogenomic integration ( PNAS 2008, PLoS One 2011, Cell 2016)- data quality improvements ( JPR2014, JASMS 2014, Bioinformatics 2015)- lipids ( Bioinformatics 2017).- intact proteins ( Nature Methods 2017), - de novo peptide identification. ( bioRxiv)
- How can we be more confident in spectral identifications?
- How can we identify spectra faster?
- Will metabolomics ever be solved?