Supported T cell receptor types [REPLACEMENT ISSUE] about scirpy

[grst]

In GitLab by @grst on Jan 24, 2020, 10:10

10x files:

We use the file filtered_contig_annotations.csv.

# Human B cell chains
(default) sturm@hochvogel Downloads % cut -f6 -d, vdj_v1_hs_pbmc3_b_filtered_contig_annotations.csv | sort | uniq -c
      1 chain
    929 IGH
    624 IGK
    506 IGL

# Human T cell chains
(default) sturm@hochvogel Downloads % cut -f6 -d, vdj_v1_hs_pbmc3_t_filtered_contig_annotations.csv | sort | uniq -c
      1 chain
     46 Multi
   4907 TRA
   5168 TRB

# Mouse B cell chains
(default) sturm@hochvogel Downloads % cut -f6 -d, vdj_v1_mm_pbmc4_b_filtered_contig_annotations.csv| sort | uniq -c 
      1 chain
   5215 IGH
   5475 IGK
   2573 IGL

# Mouse T cell chains
(default) sturm@hochvogel Downloads % cut -f6 -d, vdj_v1_mm_pbmc4_t_filtered_contig_annotations.csv| sort | uniq -c
      1 chain
      7 Multi
    761 TRA
   1301 TRB

There are indeed a bunch of barcodes that have more than 4 chains, e.g.

TTTACTGTCACCAGGC-1  True     TTTACTGTCACCAGGC-1_contig_1  True             648     TRB    TRBV19        None    TRBJ2-1   TRBC2   True         True        CASSISTDWGNEQFF              
TTTACTGTCACCAGGC-1  True     TTTACTGTCACCAGGC-1_contig_2  True             511     TRA    TRAV23/DV6    None    TRAJ58    TRAC    True         True        CAASQETSGSRLTF               
TTTACTGTCACCAGGC-1  True     TTTACTGTCACCAGGC-1_contig_3  True             521     TRB    TRBV6-5       None    TRBJ2-1   TRBC2   True         True        CASSYRTGSSYNEQFF             
TTTACTGTCACCAGGC-1  True     TTTACTGTCACCAGGC-1_contig_4  True             659     TRA    TRAV8-6       None    TRAJ6     TRAC    True         True        CAVNPGGSYIPTF                
TTTACTGTCACCAGGC-1  True     TTTACTGTCACCAGGC-1_contig_5  True             427     TRB    TRBV7-8       None    TRBJ2-1   TRBC2   True         False       CQQLRKTSYNEQFF               
TTTACTGTCACCAGGC-1  True     TTTACTGTCACCAGGC-1_contig_6  True             463     TRA    TRAV13-1      None    TRAJ4     TRAC    True         False       CSKFLFSGGYNKLIF        

But for those I checked, there are only four that are productive.
For now, I think it's fine to just use the productive chains and emit a warning that there might be more.

[grst]

In GitLab by @grst on Jan 24, 2020, 10:55

TraCeR

• Supports alpha, beta, gamma and delta. Per default, it will only look for alpha/beta
• The final output ("filtered") will contain only the two most highly expressed ones from each locus. That means max. 4 for alpha/beta. No idea how this looks like if it searches for gamma/delta as well.

Files to use:

• recombinants.txt: Output of tracer summary. Contains CDR3 sequences, V and J gene of the filtered chains.
• <CELL>/filtered_TCR_seqs/filtered_TCRs.txt. Contains V gene, J gene, CDR3 seqs and TPM of the filtered chains. Seems to be the only way to get the TPMs unfortunately.

For now, let's go for TraCeR alpha/beta only.

[grst]

In GitLab by @szabogtamas on Jan 24, 2020, 13:04

Totally agree: even four chains is a lot and we can safely assume that there shouldn't be more than four productive chains in a cell. Since we only have datasets with alpha/beta, I would also leave gamma/delta for now. We can include them later.

[grst]

In GitLab by @grst on Feb 14, 2020, 16:18

alpha/beta is fine for now. But this needs to be documented.

[grst]

In GitLab by @grst on Mar 27, 2020, 11:31

assigned to @szabogtamas