TCR-seq methods: Q&A

A study previously compared BCR-seq methods, but it only included a few methods. Of course, we can apply some findings from the TCR to the BCR. However, the BCR also has other complex information, specifically somatic hypermutations. This would require its own proper study and appropriate comparisons of BCR data.

TRA poses more challenges due to the increased number of genes involved, as compared to T-cell receptor β. However, there are protocols that exist for it.

In “Benchmarking of T cell receptor repertoire profiling methods reveals large systematic biases,” we use standardized names to facilitate discussion. All of the method identifiers can be found in the supplementary information. RACE-3 is Takara Bio’s SMART-Seq® TCR a/b Profiling Kit v1.

I will not say that in general, mPCR is more biased than RACE, because some mPCR methods performed quite well. Some RACE methods are quite efficient, others less. The same can be said for mPCR techniques.

Certainly, there is an impact from the input nucleic acid material. RNA in different populations can be highly variant, and specific bioinformatics solutions might be required to enable comparisons.

To our knowledge, there is only one commercially available T-cell line, CTLL-2. Eventually, primary cells from TCR transgenic mice with a RAGKO background, such as OT-I or OT-II, can be used.

The minimum amount required depends on the initial number of cells and the contents within each cell. Other downstream adaptations are required to prevent over-sequencing.

In our study, the lowest amount of RNA tested is 10 ng (Supplementary Figure 8). We compared the impact of a smaller quantity of RNA with a larger quantity, 100 ng. The conclusions are as follows:

• For all methods, richness was higher in large (100 ng) than small (10 ng) samples (Supplementary Figure 8A).
• Renyi diversity profiles (Supplementary Figure 8B) showed when α<2 (that is, when the diversity metric is influenced by rare clones), the diversity of small samples is less than that of larger ones. In contrast, at α≥2 (Simpson index), diversity profiles of both samples overlap. Thus, a low RNA input influences the number of rare TCR sequences detected but not the distribution of the more abundant TCRs.
• Finally, the inter-sample similarity (Supplementary Figure 8C) shows that for TRA, the similarity between 10-ng replicates was lower at the level of VJ usage and of all clonotypes, compared with that between 100-ng replicates. For TRB, the results were similar regardless of the quantity (MH>0.5). When focusing on the 1% MPC, the similarity was comparable regardless of the quantity for both TRA and TRB. These results indicated that RNA quantity, indeed, affects rare clonotype detection.

This depends on the protocols and the size of the generated amplicons. In order to detect the maximum number of T-cell clones, you must adjust the sequencing depth more than the number of cycles.

The sensitivity of the method is also very important. Based on the table, the most sensitive methods are RACE-3 and RACE-5 for the TRA chain, and mPCR-1, mPCR-3, and RACE-3 for the TRB chain.

For RACE-PCR, I suggest going for at least 250, but you could make adaptations with 150 as some of the co-authors of our study used to do.

Yes, in a previous, smaller comparison, but the results were not satisfactory.

I have no experience in BCR or B cells, but you may want to contact people such as Dr. Nina Luning Prak or Dr. Chaim Schramm.

No, we did not test for differential expression. Nevertheless, we can hypothesize that as Jurkat cells are leukemic clones, they may have an activation state and thus a higher TCR RNA expression than naive T cells, which constitute the majority of the T cells in our study.

With respect to the protocols performed by academic laboratories and service providers, the protocols have indeed been "standardized." For those carried out by our laboratory (RACE-3, RACE-6, and mPCR-3), we have done some experiments beforehand. The production of all libraries from these three methods was executed by the same person. The validation steps requested by the suppliers were performed. It is therefore possible that technical biases are linked to the efficiency of library production, but this is a bias to be considered in the choice of the method.

This is a very important question. I do not have the answer because using a unique clone may not help evaluating all potential issues during a library prep or run, or even to handle the complexity. There are several publications addressing this, but there are no gold standards yet. It depends on the nature of your repertoires and the research questions. This is a topic we are addressing in the AIRR Community—trying to identify what controls and standards are appropriate. If you are interested, please get in touch with the working group.