Several considerations, Marc. First, single-cell is a different frame of mind from classic RNAseq, where you’d want triplicates – or more - from the start to validly look for differentially expressed genes. Here, each cell is an independent object. So you might be asking “how do my 120 tetramer+ cells compare to the 5,324 polyclonal cells in the same run”, completely different stats. So its not as if you needed to plan up front in terms of 3 replicates for each time point. On the other hand, any conclusion from a single-cell experiment needs to be observed in [at least] 2 different experiments (basic reproducibility). But quite often, the “conclusion” (or suggestion thereof) will come from the first experiment, and what needs to be replicated will only become apparent then (and perhaps fine-tune the experimental design) . So, pretty often, strategy is to do one run, see what you get, adjust the repeat experiment (e.g. more of some cells, fewer of others, etc). For a study of tetramer+ cells, though, my sense is that you’d want them to come from several different mice, up-front. Hash-tagging would be a great way to flag which cells come from which mouse. To see how the repertoires compare between mice, whether the tet+ cells acquire the same characteristics in different animals, etc.. Unless you need to pool 10 donors just to get one sample. cb

For thymocytes, is definitely on to run balanced sampling, otherwise data will be overwhelmed by pretty boring DPs. What is currently on the books is to rebalance by over-sampling double-negatives and intermediate/mature TCRhi cells. But there are probably sharper ways to rebalance (especially for the gdT side of things) And yes, having some relation to TCR strength-of-signal would definitely be desirable. Not sure that a single-chain TCR transgenic would give us that handle (variability in the second chain would likely create a wide range of affinities for thymic MHC-peptide complexes). But selection by tetramer staining (that one perhaps coupled with a single-chain Tg), or focusing on super-antigen reactive T cells, might get us there.

Right, let's include those. Anyone have coordinated sets of tetramers and immunizations that could be brought to bear? John A,, do you think this is something that the Tetramer Core would be interested in participating in, on a broad scale? Could be an exciting subproject...

Interesting suggestion, raises the question of KO mice within immgenT. Since the scope of the project is to profile "all the states that T cells can adopt", transcription factor knockouts that artificially perturb the transcriptional regulatory network would not belong (don't think that we should include Tbet knockouts, for instance). But here the situation is different. If I understand what you are suggesting, would not aim to analyze the artificifial "wannabe Tregs*" , but the FoxP3-negative conventional T cells that respond in the autoimmune context of a Treg deficiency (in addition, it is a model for the human IPEX disease) So let's do it ! * of Rudensky/Chatila fame (messed up cells that have many characteristics of Tregs but lack FoxP3)

Would be most grateful for specific suggestions

Yes, this program will be trickier than conventional RNAseq, for which frozen lysates travel easily. Principally, there will be two main immgenT sites. One in San Diego and one in Boston, who will use the same reagents and strict SOPs. From there, several scenarios are possible, depending on the actual types of samples: for "easy" cases, simply send the mouse protocol/mutants to one of the two sites, and the preparations will be performed by the core team; if more complex, investigators go to Boston or San Diego to prep and run tricky samples*. There is also the outside possibility of having additional sites runs the cell preparations and encapsulation (not preferred, in order to limit experimental drift) * this will happen for "analysis jamborees"

Cautiously... ImmGen has long had 'In Vivo Veritas" as a motto, and the general concept of the immgenT project is very much in that vein. That being said, a case could be made that some of the extreme states reached in vitro by stimulating T cells under various cytokine cocktails could serve as useful references. So let's talk...

Indeed. And ImmGen means are not infinite. The matrix of cells and challenged conditions is very large, and we could certainly not dream of encompassing, in a single-cell project, all possible permutations of locations, challenges, times, etc. But the goal is to chart all of the phenotypes that a T cell can be pushed into, from naive resting cells to the most extreme states of differentiation that they can adopt. Hence, the project is not conceived as a few highly detailed analyses, rather a broad yet sparse survey, selectively choosing locations and conditions for maximum informativity. Some dose/response or time-course will be done, for sure. But not everywhere.