AIRRC5 - Analyzing the space of TCRs using optimal transport (F. Matsen)
120 بار بازدید -
4 سال پیش
-
Session from the "AIRR Community
Session from the "AIRR Community Meeting V – Zooming in to the AIRR Community!" (December 8-10, 2020)
The meeting covers progress reports from the AIRR-C Working Groups & Sub-committees, two scientific sessions (including four short talks chosen from the posters), two interactive poster sessions and four live software tool demonstrations.
https://www.antibodysociety.org/the-a...
Analyzing the space of TCRs using optimal transport
Frederick Matsen
Professor, Public Health Sciences Division, Fred Hutchinson Cancer Research Center
Immune receptors implicitly contain rich information about immune exposures, cell types, and host genetics. However, computational algorithms are required to gain insight from these data. One can imagine constructing an abstract "space" in which T cell receptor (TCR) sequences with similar properties are nearby one another. Then, immune repertoire comparison becomes a task of comparing "regions" of space that are more densely occupied in one repertoire but not in another. In this talk I will describe our first steps toward realizing such comparison. We use a space of TCR sequences defined by the TCRdist distance. Rather than using dimension reduction on that space, we compare repertoires directly in that space using the methods of "optimal transport." I will also contrast this technique with other useful means of comparing repertoires, including summary statistics and parametric probabilistic models.
The meeting covers progress reports from the AIRR-C Working Groups & Sub-committees, two scientific sessions (including four short talks chosen from the posters), two interactive poster sessions and four live software tool demonstrations.
https://www.antibodysociety.org/the-a...
Analyzing the space of TCRs using optimal transport
Frederick Matsen
Professor, Public Health Sciences Division, Fred Hutchinson Cancer Research Center
Immune receptors implicitly contain rich information about immune exposures, cell types, and host genetics. However, computational algorithms are required to gain insight from these data. One can imagine constructing an abstract "space" in which T cell receptor (TCR) sequences with similar properties are nearby one another. Then, immune repertoire comparison becomes a task of comparing "regions" of space that are more densely occupied in one repertoire but not in another. In this talk I will describe our first steps toward realizing such comparison. We use a space of TCR sequences defined by the TCRdist distance. Rather than using dimension reduction on that space, we compare repertoires directly in that space using the methods of "optimal transport." I will also contrast this technique with other useful means of comparing repertoires, including summary statistics and parametric probabilistic models.
4 سال پیش
در تاریخ 1399/09/27 منتشر شده
است.
120
بـار بازدید شده