I am trying to run RosettaAntibody3 to build an Fv hhomology model from an input sequence. I am following the documentation form here.
I am using the command:
antibody.static.linuxgccrelease \
-database $ROSETTA_DATABASE \
-fasta /mnt/data/input/myinput.fasta \
-nstruct 1 \
-out:path:all /mnt/data/output/output-graft/ \
-detect_disulf false
which fails fairly quickly with the error:
protocols.antibody.grafting: Path to REGEX file is: /databases/rosetta/protocol_data/antibody/cdr_regex.txt
protocols.antibody.grafting: The detected H1 REGEX is: C[A-Z]{1,16}(W)(I|V|F|Y|A|M|L|N|G)(R|K|Q|V|N|C|G)(Q|K|H|E|L|R)
protocols.antibody.grafting: The detected H3 REGEX is: C[A-Z]{1,33}(W)(G|A|C)[A-Z]{1,2}(Q|S|G|R)
protocols.antibody.grafting: The detected L1 REGEX is: C[A-Z]{1,17}(WYL|WLQ|WFQ|WYQ|WYH|WVQ|WVR|WWQ|WVK|WYR|WLL|WFL|WVF|WIQ|WYR|WNQ|WHL|WHQ|WYM|WYY)
protocols.antibody.grafting: The detected L3 REGEX is: C[A-Z]{1,15}(L|F|V|S)G[A-Z](G|Y)
protocols.antibody.grafting: H1 detected: GFTLEYYAIG (10 residues at positions: 25 to 35)
H3 detection: failed
[ ERROR ]: Caught exception:
File: src/protocols/antibody/grafting/regex_based_cdr_detection.cc:100
H3 detection: failed. Pattern:C[A-Z]{1,33}(W)(G|A|C)[A-Z]{1,2}(Q|S|G|R)
I know this is a valid sequence - as the antibody itself is a real and functional antibody.
Does anyone know why RosettaAntibody is failing to detect H3 as is? Is there a way to circumvent this problem?
As a hack, I changed the cdr_regex.txt to replace the H3 line with `H3 "<MyKnownH3Sequence>"`, and it still failed.
Category:
Post Situation:
I was able to get some help on this via the back channels.
One way to circumvent this issue is to supply the positions of the CDRs manually using the -antibody:json_cdr <filename> flag with the format:
Passing this file skipped the auto CDR detection, and all worked well. Solved.