junction_aa contains extra AA before conserved C residue for TCR alpha

[Scott Christley]

I've been in the process of updating the apps on VDJServer, including moving to IgBlast 1.8.0 and more recent pRESTO and ChangeO. I'm using default branch so that I get the AIRR format.

$ hg summary
parent: 665:ac5ceba07a80 tip
 Merged kleinstein/changeo into default
branch: airr-handle-only
commit: 35 unknown (new branch)
update: (current)
phases: 5 draft

Anyways, I processing a TCR alpha data set and noticed extra AAs before the conserved C residue. I don't process TCR alpha too often, so I don't know if this was an issue with earlier version of ChangeO or not. Here's a couple sequences as examples:

CCTATCCCCTGTGTGCCTTGGCAGTCTCAGCAGGTCTTCAGTTGCTTATGAAGGTTTTCTCAAGTACGGAAATAAACGAAGGACAAGGATTCACTGTCCTACTGAACAAGAAAGACAAACAACTCTCTCTGAACCTCACAGCTGCCCATCCTGGGGACTCAGCCGTGTACTTCTGCGCAGCCCCCGGGACTGGAGGCTATAAAGTGGTCTTTGGAAGTGGGACTCGATTGCTGGTAAGCCCTGACATCCAGAACCCGGAACCTGCTGT
CCTATCCCCTGTGTGCCTTGGCAGTCTCAGGAGAAGGTCCACAACTCCTCTTTAGAGCCTCAAGGGACAAAGAGAAAGGAAGCAGCAGAGGTTTTGAAGCTACATATGATAAAGGGACCACCTCCTTCCACTTGCGGAAAGCTTCAGTGCAAGAGTCAGACTCGGCTGTGTACTACTGTGCTCTGAGTGGTAGGCACTATGGAAATGAGAAAATAACTTTTGGGGCTGGAACCAAACTCACCATTAAACCCAACATCCAGAACCCGGAACCTGCTGT

with junction_nt coming out as

TACTTCTGCGCAGCCCCCGGGACTGGAGGCTATAAAGTGGTCTTT
TACTACTGTGCTCTGAGTGGTAGGCACTATGGAAATGAGAAAATAACTTTT

and junction_aa coming out as

YFCAAPGTGGYKVVF
YYCALSGRHYGNEKITF

IgBlast gives the CDR3 as this

AAPGTGGYKVV
ALSGRHYGNEKIT 

[Scott Christley]

My version of ChangeO might be a bit older than I expect, so I'm gonna update it and try again.

[Scott Christley]

This is mouse.

[Scott Christley]

Okay, I think I'm at the most current and I get the same result.

$ hg summary
parent: 696:35caac882b75 tip
 Added AIRR format support to ConverDb-genbank.
branch: default
commit: 28 unknown (clean)
update: (current)

[Jason Vander Heiden]

Okay. I'll take a look at this soon (hopefully today). This would mean the IMGT numbering isn't correct, because the junction start position is fixed at 309.

[Jason Vander Heiden]

So this is caused because amino acid 104 is actually in position 106 in these TCRA genes.

IMGT alignment:

                                    1               5               8A      10                  

1                                   ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... 
                                     G   E   Q   V   E   Q   R   P       P   H   L   S   V   R  
AC163653 Musmus TRAV3D-3*02 F       ggc gag cag gtg gag cag cgc cct ... cct cac ctg agt gtc cgg 

                                    75                  80                  84A 85              
                                     G   F   T   V   L   L   N   K   K   D       K   Q   L   S  
1                                   gga ttc act gtc cta ctg aac aag aaa gac ... aaa caa ctc tct

Note, that TRAV3D-3*02 has both a position 8 and an 8A (9) as well as an 84 (85) and 84A (86). That's what's screwing it up. The IMGT-gapping itself is fine, w.r.t. the IMGT germlines, it's just that these particular germlines do not adhere to the IMGT numbering scheme. :/

This is an instance where it's probably just easier to just pass the IgBLAST CDR3 to JUNCTION (with +/-3) rather than add a fix.

I might fix it anyway by adding a motif search for the conserved residue instead of relying on the IMGT numbering, but... probably more trouble than it's worth.

[Jason Vander Heiden]

Okay, I updated the parser to use IgBLAST's CDR3 for determining the junction. Only works for IgBLAST v1.7+. Falls back to the old IMGT-gapping method for older versions.

[Jason Vander Heiden]

Hey @schristley, can I close this? Is this working?

[Scott Christley]

Test jobs are running now so should know soon.

[Scott Christley]

CDR3 looks to be coming out correctly for my data sets.