Gene fusion curation (Room 308)

[ahwagner]

The VICC gene fusions project, in partnership with representatives from ClinGen Somatic, CGC, and the CAP/ACMG Cytogenomics Committee, has developed a draft specification for the representation of gene fusions: fusions.cancervariants.org.

I propose that a survey of existing categorical fusion concepts in the CIViC knowledgebase is evaluated under this framework and structured information collected to improve computable interpretation and search of CIViC fusion variants.

Depending on the expertise of participants, available curation tools to enable this exercise include hypothes.is or Jupyter notebooks leveraging the FUSOR package.

[ahwagner]

Useful tooling: take chimeric fusion sequences and generate an assayed fusion under the specification guidelines

Example fusion sequence from this page: https://ccsm.uth.edu/FusionGDB/gene_search_result.cgi?page=page&type=quick_search&quick_search=32394

From this, find representative transcript fusion, annotated with reading frame prediction, preserved domains, etc. and match to curated evidence

[ahwagner]

Our workflow: https://fusions.cancervariants.org/en/draft-1/workflows.html

Our example exercise is capturing the ALK Fusion page from CIViC: https://civicdb.org/variants/499/summary

Our sheet for curating these variants: https://docs.google.com/spreadsheets/d/1WCgrDAqMKj0AOk45uGEJ2koS3T4SZ_nnryHKqGB9pcg/edit?usp=sharing

[ahwagner]

A future extension for this work: how do we describe a group of genes with known oncogenic properties, for more precise representation of concepts such as reg_e@IGH::v?

[ahwagner]

@bpitel12 ran a tutorial on the St. Jude ProteinPaint fusion visualization tool.

[ahwagner]

Group discussion point: workflow should get more concretely to the point of the chimeric protein being of relevance.

Generalize the critical domains to also consider regulatory (non-translated) domains within genes

IGH::MYC needs to be added to CIViC! @jsaliba10 reminded us that anyone in the community can do this. 😄

[ahwagner]

During the curation exercise, it was found that the EWSR1::FLI1 structure was defined by CIViC as:

In approximately 90% of Ewing sarcoma/primitive neuroectodermal tumor (ES/PNET) a translocation t(11;22)(q24; q12) is identified, fusing the 5’ exons (containing the transactivation domain, at least exon 1-7) of EWSR1, with the 3’ exons (from exon 9, coding the DNA binding domain) of FLI1.

These exon structures are correct, but there is no annotated transactivation domain for EWSR1; as described in https://acsjournals.onlinelibrary.wiley.com/doi/full/10.1002/cncy.22239:

The N-terminus (exons 1-7) mediates transcriptional activation through degenerate repeats of the SYGQ motif, whereas the C-terminus contains an 87-amino-acid RNA-recognition motif encoded by exons 11 through 13.4 Although variable, the majority (80%) of the breakpoints in EWSR1 occur in intron 7 or 8, resulting in fusion of the EWSR1 N-terminus to heterologous DNA-binding domains of its partners.

This helped us identify the need to define categorical fusions by minimal included exons in addition to annotation of critical protein domains.

[ahwagner]

Add information about observed fusion junctions to the "valid junction range" concept discussed above.

[malachig]

Summary of activities presented on behalf of the group by Beth Pitel and Alex Wagner

[bpitel12]

Fusion Curation_CIViC 2022.pptx

[skr1]

would love to be the part of working group for protein domain annotation of novel fusion and perhaps protein structure prediction as well to contribute towards their functional evidence.

[skr1]

Please indicate the name of NIH consortium working on such efforts.

[malachig]

Resolving in preparation for the 2023 hackathon/jamboree.