What it does

Assuming you have defined the familial relationships between samples when loading your VCF into GEMINI, you can use this tool to identify candidate genes and variants that explain the inheritance pattern of a phenotype of interest.

Inheritance pattern detection rules

Autosomal recessive

Criteria:

• all affected must be hom_alt
• [affected] no unaffected can be hom_alt (can be unknown)
• [default] if parents exist they must be unaffected and het for all affected kids
• [default] if there are no affecteds that have a parent, a warning is issued.

If --lenient is specified, the 2 criteria prefixed with “[default]” are not applied.

If --allow-unaffected is specified, the criterion prefixed with “[affected]” is not enforced.

Autosomal dominant

Criteria:

• All affecteds must be het
• [affected] No unaffected can be het or homalt (can be unknown)
• de_novo mutations are not auto_dom (at least not in the first generation)
• At least 1 affected must have 1 affected parent (or have no parents).
• If no affected has a parent, a warning is issued.
• [default] All affecteds must have parents with known phenotype.
• [default] All affected kids must have at least 1 affected parent

If --lenient is specified, the criteria prefixed with “[default]” are not enforced.

If --allow-unaffected is specified, the criterion prefixed with “[affected]” is not enforced.

Note that, for autosomal dominant, --lenient allows singleton affecteds to be used to meet the --min-kindreds requirement if they are HET.

If there is incomplete penetrance in the kindred (unaffected obligate carriers), these individuals currently must be coded as having unknown phenotype or as being affected.

X-linked recessive

Criteria:

• Affected females must be HOM_ALT
• Unaffected females are HET or HOM_REF
• Affected males are not HOM_REF
• Unaffected males are HOM_REF

Note: Pseudo-autosomal regions are not accounted for by the tool.

X-linked dominant

Criteria:

• Affected males are HET or HOM_ALT
• Affected females must be HET
• Unaffecteds must be HOM_REF
• girls of affected dad must be affected
• boys of affected dad must be unaffected
• mothers of affected males must be het (and affected)
• at least 1 parent of affected females must be het (and affected).

Note: Pseudo-autosomal regions are not accounted for by the tool.

De-novo mutations

Criteria:

• all affected must be het
• [affected] all unaffected must be homref or homalt
• at least 1 affected kid must have unaffected parents
• [default] if an affected has affected parents, it’s not de_novo
• [default] all affected kids must have unaffected (or no) parents
• [default] warning if none of the affected samples have parents.

The last 3 items, prefixed with [default] can be turned off with --lenient.

If --allow-unaffected is specified, then the criterion prefixed [affected] is not enforced.

X-linked de-novo mutations

Criteria:

• affected female child must be het
• affected male child must be hom_alt (or het)
• parents should be unaffected and hom_ref

Note: Pseudo-autosomal regions are not accounted for by the tool.

Compound heterozygosity

Unlike canonical recessive sites where the same recessive allele is inherited from both parents at the same site in the gene, compound heterozygosity occurs when the individual’s phenotype is caused by two heterozygous recessive alleles at different sites in a particular gene.

To detect compound heterozygosity, the tool looks for two heterozygous variants impacting the same gene at different loci. The complicating factor is that this is a case of recessive inheritance and as such, we must also require that the consequential alleles at each heterozygous site were inherited on different chromosomes (one from each parent). Hence, where possible, the tool will phase by transmission.

Criteria (default):

• All affected individuals must be heterozygous at both sites.

• No unaffected can be homozygous alterate at either site.

• Neither parent of an affected sample can be homozygous reference at both sites.

• If any unphased-unaffected is het at both sites, the site will be given lower priority.

• No phased-unaffected can be heterozygous at both sites.

  1. --allow-unaffected keeps sites where a phased unaffected shares the het-pair
  2. unphased, unaffected that share the het pair are counted and reported for each candidate pair.

• Candidates where an affected from the same family does NOT share the same het pair are removed.

• Sites are automatically phased by transmission when parents are present in order to remove false positive candidates.

  If data from one or both parents are unavailable and the child’s data was not phased prior to loading into GEMINI, all comp_het variant pairs will automatically be given at most priority == 2. If there’s only a single parent and both the parent and the affected are HET at both sites, the candidate will have priority 3.

Criteria (--pattern-only):

• Kid must be HET at both sites.
• Kid must have alts on different chromosomes.
• Neither parent can be HOM_ALT at either site.
• If either parent is phased at both sites and matches the kid, it’s excluded.
• When the above criteria are met, and both parents and kid are phased or parents are HET at different sites, the priority is 1.
• If either parent is HET at both sites, priority is reduced.
• If both parents are not phased, the priority is 2.
• For every parent that’s a het at both sites, the priority is incremented by 1.
• The priority in a family is the minimum found among all kids.

Violation of Mendelian laws

The tool can be used to detect the following kinds of non-Mendelian patterns:

• loss of heterozygosity (LOH) events
• de-novo mutations
• implausible de-novo mutations
• potential cases of uniparental disomy

Criteria:

• LOH: child and one parent are opposite homozygotes; other parent is HET
• plausible de novo: kid is het. parents are same homozygotes
• implausible de novo: kid is homozygote. parents are same homozygotes and opposite to kid.
• uniparental disomy: parents are opposite homozygotes; kid is homozygote