Memory CD4+ T cell receptor repertoire data mining as a tool for identifying cytomegalovirus serostatus

Abstract

Pathogens of past and current infections have been identified directly by means of PCR or indirectly by measuring a specific immune response (e.g., antibody titration). Using a novel approach, Emerson and colleagues showed that the cytomegalovirus serostatus can also be accurately determined by using a T cell receptor repertoire data mining approach. In this study, we have sequenced the CD4 memory T cell receptor repertoire of a Belgian cohort with known cytomegalovirus serostatus. A random forest classifier was trained on the CMV specific T cell receptor repertoire signature and used to classify individuals in the Belgian cohort. This study shows that the novel approach can be reliably replicated with an equivalent performance as that reported by Emerson and colleagues. Additionally, it provides evidence that the T cell receptor repertoire signature is to a large extent present in the CD4 memory repertoire.

Deep generative models for T cell receptor protein sequences

“…To balance these evaluative and generative objectives, we fixed β to be 0.75. This choice was confirmed by running the same analysis on the data of Emerson et al (2013) (Figure 9—figure supplement 1) and De Neuter et al (2019) (Figure 9—figure supplement 2), both of which yielded similar results.…”

“…To test the models’ compliance with the rules of VDJ recombination, we used the data of De Neuter et al (2019), which consists of TCR β sequences from 33 subjects, as follows. We randomly split the data so that the repertoires of 22 subjects were used for training, and the remaining 11 subjects’ repertoires were used for testing (with one repertoire used for each subject).…”

“…To evaluate out-of-sample probability estimation, we used the De Neuter et al (2019) data as in the previous section to evaluate PVAE under the model rather than true0P. If the VAE were regurgitating training sequences, it should consistently assign higher PVAE to sequences it generates compared to held-out test sequences.…”

“…

10.7554/eLife.46935.007Figure 5—figure supplement 1.Nearest neighbor Levenshtein distributions on the De Neuter et al (2019) data set.Nearest neighbor Levenshtein distance distributions for simulated sequences and test repertoire sequences. Each of the divergences in Figure 4 calculate a divergence between one of the colored lines (a simulated collection of sequences) and one of the gray lines (test repertoire sequences).

…”

“…

10.7554/eLife.46935.015Figure 9—figure supplement 2.Summary statistic divergences for each on the De Neuter et al (2019).Analysis identical to that in Figure 7.

…”