Dr. Desaire, Dr. Go, and four researchers from the Desaire group seated outside of the Multidisciplinary Research Building during the summer of 2021.

Our Mission

We apply our core competencies, mass spectrometry and machine learning, to develop tools and biomarkers that will improve human health and address health disparities in underserved populations.

Our Research

Chronic Kidney Disease

  • Urine
  • Glycomics
  • Direct Infusion ESI-MS

a glycoprotein sample in a vial showing an arrow pointing to the different glycans that can be recovered from ita drawing of the kidneysan example of a label-free mass spectrum and the resulting machine learning classification plot via the Aristotle Classifier (different samples cluster in opposite quadrants of the plot)

 

Alzheimer's Disease

  • Plasma
  • Proteomics
  • LC-MS/MS

an example of blood plasma in a vial, showing the red and white blood cells that are present in the layers beneaththe structural representation of adiponectin proteina comparison between healthy brain and Alzheimer's brain in which the Alzheimer's brain shows significantly more deterioration and shrinkage

 

Nutritional Insufficiency

  • Fingerprints
  • Metabolomics/Lipidomics
  • Flow Injection ESI-MS

a cartoon drawing of a fingerprintthe output of the Aristotle classifier, in which opposite quadrants share the same color; samples cluster in the top left and bottom right quadrants, while only a few samples cluster in the top right and bottom leftcartoon outline of a mother nursing her child

 

                                              chemical line structure of a generic triglyceride

chemical line structure of a generic wax esterchemical line structure of a generic fatty acid

Latent fingerprints consist of water, proteins, amino acids, lipids, and salts. Lipids are particularly relevant analytes because they are known to vary in a variety of diseases of interest (neurodegeneration, metabolic disorders, autism). Most skin-surface lipids originate from sebaceous gland secretion. We study multiple lipid classes present in human sebum, including triglycerides, fatty acids, and wax esters.

a typical mass spectrum of a fingerprint, showing peaks most evident in the 400 - 600 m/z range and the 750-950 m/z range

MS Analysis of Latent Fingerprints

  • Determination of changes in lipid profiles
  • High-throughput
  • Simple sample prep
  • Noninvasive sample collection
  • Relatively inexpensive

a picture of a fingerprint (groomed) and an arrow pointing to the resulting lipid (triglyceride) line structure that is produced after the liquid-liquid extraction stepan example of the MS data acquired from the fingerprint sample, with an arrow pointing to the computer which indicates machine learning of differences in lipid profiles, and another arrow pointing to a fingerprint with a question mark, which indicates the determination of an unknown biological state

For disease classification, the whole lipidomic profile can provide a more complete picture than a few lipids can by themselves.

an example of a painting of someone outside holding an umbrella, in the first example many portions of the painting are blank and it is difficult to understand the full picture, but in the second all of the pieces are present and the complete picture is clear

The acquisition of a large MS dataset can introduce sources of systematic variation, such as batch effects, which can mask the biological variation of interest. Normalization is a way to minimize these effects of systematic variation. The effectiveness of different normalization strategies is often data-dependent, so it is necessary to assess the performance of different strategies.

an example of the resulting PCA plot from fingerprint samples collected on the MS over multiple different days, in the first plot the samples of different days are clustered according to day, in the second plot which is normalized the clusters are all on top of one another and overlapping, suggesting the data is comparable

We are grateful to the National Institutes of Health for supporting our work!

To learn more, contact:

Prof. Heather Desaire

hdesaire@ku.edu

220D MRB