TCR/BCR enhancements

[Zethson]

• [ ] Review by BS
• [x] Tessa etc output in multimodal integration way way too verbose. Silence it
• [ ] Update https://www.sc-best-practices.org/air_repertoire/ir_profiling.html#immune-receptor-sequencing for latest manuscript changes -> no one uses these ancient tools -> point to modern ones

[zhouzhendiao]

May I ask which tools are too old to use? Have they be updated yet?

[Zethson]

@zhouzhendiao the corresponding section in the not yet published manuscript:

AIRR sequences can be deciphered with V(D)J-sequencing followed by alignments and chain pairing. Although no benchmarks exist for TCR sequence reconstruction, MiXCR and TRUST4 are frequently used. BALDR, BASIC and BraCer were shown to robustly recover BCR sequences but are not maintained anymore. We therefore encourage analysts to also consider the more recent MiXCR and TRUST4 for BCR sequences.

[zhouzhendiao]

@zhouzhendiao the corresponding section in the not yet published manuscript:

AIRR sequences can be deciphered with V(D)J-sequencing followed by alignments and chain pairing. Although no benchmarks exist for TCR sequence reconstruction, MiXCR and TRUST4 are frequently used. BALDR, BASIC and BraCer were shown to robustly recover BCR sequences but are not maintained anymore. We therefore encourage analysts to also consider the more recent MiXCR and TRUST4 for BCR sequences.

Thanks for replying~

[zktuong]

TraCer/BraCer is still being used for plate-based single-cell data. Although maybe not as up to date compared to other tools that do reconstruction, they still perform well (doesn’t artificially insert V genes into contigs for example which TRUST4 seems to do)

[Zethson]

Thanks @zktuong - would you be interested in amending the content to ensure that it's factually correct?

[zktuong]

I'll try when I get the opportunity to sit down and write something to summarise the results of the benchmarking paper by Andreani 2022 paper, as well as some of the nuance things we found in Suo 2023.