Home Cytometry History Individual Histories James Leary
James Leary
- What do you feel was your major contribution to the field of Cytometry?
- I invented some of the first basic technology for high-speed cell sorting and many of the basic methods for rare-event analysis. Since going faster alone was not the solution, this led me into how to classify cells, including rare cells, with complex mulitvariate statistical classifiers in real time to sort those cells. All of this resulted in many publications and 5 issued US Patents.
- What drove you to this achievement?
- At the time I was working on trying to isolate very rare human fetal cells from maternal blood. The technology necessary to accomplish this simply did not exist.
- Why was it that your team was able to do it? Was it a special skill? Sudden insight?
- I like to think that we had did what we did because we did not have enough research support to do a brute-force solution, but we did have enough support to do a fairly creative solution.
- Was someone else’s work or influence fundamental in driving your work?
- At the time I felt that I was a young Assistant Professor with a small amount of resources competing against a vast army of people at Lawrence Livermore Lab (who were working on high-speed chromosome sorting). The competition was always friendly, but it forced me to think creatively about the problem, since creativity was a significant portion of my total resources at the time!
- How do you think your work impacted the field of Cytometry?
- I like to think that I took rare-event analysis from the impossible to the possible. When I started this work most people did not think that it could be done. Others could not understand why this was worth doing. It makes me feel good to see so many people doing rare cell analysis now and now appreciating its importance to so many fields of biology and medicine.
Overview of a 6-color high-speed flow cytometer with a single-layer neural network architecture that allows real-time classification/sort decisions on the basis of multivariate statistical functions. For high recovery of cells sorted at high speeds and for efficient single-cell recovery, we often sort cells of interest straight-ahead, achieving sort recoveries of more than 95 percent even at the single-cell level at rare cell frequencies as low as 10-6. The multivariate statistical functions calculated in real-time provide for statistically-based sort boundaries and a measure of the degree of misclassification.
Leary, J.F. “Ultra High Speed Cell Sorting” Cytometry: Part A 67A:76-85 (2005).
The Theoretical architecture of an 8-parameter high-speed cell sorter capable of making real-time multivariate statistical sort decisions (in this case the first three discriminant functions d1, d2 and d3. This architecture is similar to that of a single-layer neural network where the weighting coefficients can be used in a wide variety of ways to represent not only neural networks but also complex multivariate statistical functions such as principal components, discriminant functions and logistic regression functions. Even non-linear functions can be approximated using generalized additive functions.
Leary, J.F. “Ultra High Speed Cell Sorting” Cytometry: Part A 67A:76-85 (2005).
- James F. Leary
- SVM Professor of Nanomedicine
- Professor of Basic Medical Sciences and Biomedical Engineering
- Purdue University West Lafayette, Indiana