how to handle gene cluster deletion in PHI-Canto?

[CuzickA]

PHI-base currently has an evidence code 'gene deletion:cluster'.

Example 1
There is only 1 PHI-base entry using this evidence code! PMID:28288207
http://www.phi-base.org/searchFacet.htm?queryTerm=28288207

PHI:7041 Gene name T6SS (type six secretion system) which is actually 37Kb DNA fragment encoding a putative T6SS consisting of 7 genes. The given phenotype is 'reduced virulence'. It has not been classed as a multiple mutation and the 7 genes have not been documented.
PHI:7042 Gene 'KatN' is an effector secreted into the host via the pathogen T6SS and is given phenotype 'effector'.

How would PHI-canto capture the data for PHI:7041 the T6SS?

7 single allele deletions merged to create a multi allele?

Problem is that this would look like 7 genes individually deleted rather than deleted as a contiguous region.

Example 2
http://www.phi-base.org/searchFacet.htm?queryTerm=27096607
PHI:6195 gene 'PAI-536' with evidence code 'gene mutation' and phenotype 'reduced virulence' has a note in the comment that this mutation has resulted in the deletion of 7 pathogenicity islands. No individual gene names have been curated.

[CuzickA]

Other questions to think about

1. in a cluster deletion, is the whole multi allele genotype assessed for a given phenotype?
2. are there examples where there the predicted phenotype of only a subset of the deleted genes assessed? In this case are the remaining deleted genes considered background?

Ideas for PHI-canto
a) capture information in a comment or text field on the genotype pombase/canto#1599 (comment)
b) create a new type of allele, where the cluster is specified as a field (e.g. using the start and end of a gene range)
c) add a tick box stating that ‘this deletion was created by deleting a region'
d) specify all non-annotated cluster genes in the background

[CuzickA]

just to note- UniProt id is required for PHI-Canto gene list and genotype creation. Therefore above examples would not have been able to be entered as no gene/uniprot was specified.

[ValWood]

Yep, Flybase would need this too.

[CuzickA]

are there examples where there the predicted phenotype of only a subset of the deleted genes assessed? In this case are the remaining deleted genes considered background?

Discussed this on today's call and decided that there would be examples like this in the publications that we want to curate.

Workaround: if the paper is about the phenotype of a specific gene in the region, annotate the gene, put the other genes in the background, and describe the number of genes deleted at the start (up and downstream) in the nucleotide deletion field (using the Nucleotide Deletion allele type).

[CuzickA]

I need to make a note about the above workaround option in the genotype help pages