Skip to content

Questions about inheritanceModes, MOI scoring, and filtering behavior #643

Description

@federedef

Hello,

First of all, thank you very much for the development of Exomiser. I have been using the tool for several months now and it has been extremely useful. However, while working with it, I have encountered some difficulties understanding certain aspects of the final prioritization results.

Before opening this issue, I carefully read the official documentation (including the development version), several GitHub issues, configuration examples, and some of the Exomiser-related publications describing its different modules and scoring strategies.

For context, I am using Exomiser on single-patient analyses only:

one VCF file
one Phenopacket
no PED file
no family segregation analysis

Even after reading the documentation, I still have several questions that I hope you may help clarify.

  1. How is the inheritance mode (AD/AR/etc.) determined?

I do not fully understand how Exomiser determines whether a variant is evaluated under an autosomal dominant (AD) or autosomal recessive (AR) model.

In some cases this is understandable because the variant is associated with a known disease already annotated with a specific inheritance pattern. However, how is this handled when there is no disease association available?

  1. Role of inheritance mode in the final Exomiser score

From the publications and documentation, I understand that the final Exomiser score is obtained through a logistic combination of:

phenotype score
variant score

However, I would like to confirm whether the Mode Of Inheritance (MOI) itself contributes directly to the final Exomiser score, or whether it is only used as a filtering/grouping mechanism before the final scoring step.

In other words:

does the MOI influence the score mathematically?
or does it only determine which variants are considered compatible candidates?

  1. Questions regarding inheritanceModes configuration

In the default configuration, the following section appears:

inheritanceModes:
AUTOSOMAL_RECESSIVE_COMP_HET: 2.0
MITOCHONDRIAL: 0.2
AUTOSOMAL_RECESSIVE_HOM_ALT: 0.1
X_DOMINANT: 0.1
AUTOSOMAL_DOMINANT: 0.1
X_RECESSIVE_COMP_HET: 2.0
X_RECESSIVE_HOM_ALT: 0.1

Along with these comments:

\ In cases where you do not want any cut-offs applied an empty map should be used e.g. inheritanceModes: {}
\ These are the default settings, with values representing the maximum minor allele frequency in percent (%) permitted for an
\ allele to be considered as a causative candidate under that mode of inheritance.
\ If you just want to analyse a sample under a single inheritance mode, delete/comment-out the others.
\ For AUTOSOMAL_RECESSIVE or X_RECESSIVE ensure both relevant HOM_ALT and COMP_HET modes are present.

What I do not fully understand is the following.

My current interpretation is that inheritanceModes mainly defines MAF thresholds for each inheritance model rather than enabling/disabling inheritance analysis itself.

However:

what happens if all inheritance modes are commented out?
what happens if inheritanceModes: {} is used?

Does this mean:

no inheritance filtering is applied at all?
all inheritance models remain active but without MAF cutoffs?
or alternatively that no variants are retained?

I ask this because the comments seem somewhat ambiguous:

on one hand, they suggest that commenting out inheritance modes disables those analyses
on the other hand, they suggest that an empty map removes cutoffs globally.

  1. Difference between inheritanceModes, frequencyFilter, and inheritanceFilter

I also noticed the presence of other filtering-related parameters such as:

frequencyFilter:
maxFrequency: 2.0

and:

inheritanceFilter: {}

This raises additional questions:

how does frequencyFilter.maxFrequency differ from the thresholds already defined inside inheritanceModes?
is inheritanceFilter functionally equivalent to enabling/disabling inheritance modes?
should inheritanceModes mainly be used only when a PED/family analysis is available?

I suspect I may be misunderstanding the interaction between these parameters, especially in single-sample analyses.

Thank you very much for your time and for the development of this excellent software.

And apologies if some of the questions are not perfectly formulated, I have tried to read as much documentation and previous discussion as possible before posting this issue, but I still have these remaining conceptual doubts.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions