Fish Mitogenome Curation • MitoPilot

Automated curation and validation of mitogenome annotations is a central goal of MitoPilot. When samples are processed through the Annotation Module, annotations are first generated using a combination of MITOS2 (for rRNA and PCG annotation) and tRNAscan-SE (for tRNA annotation). Following the initial annotation, samples are proceessed through an automated curations process where modifications are made in attempt to bring the annotations closer to a state that will be considered acceptable for publication by NCBI GenBank. This automated curation process is based on 2 important inputs to the pipeline, a high-quality reference database of protein coding genes, and a set of user configurable parameters.

Reference Databases

MitoPilot expects the curation reference database files to be in the form of Blastp database files, one for each gene. By default, the pipeline will use the same database files used by Mitos2 during the initial annotation process. This can be modified using the ref_db property of the Curation Parameters (see below). The default value of ref_db, (ref_db = list(default = "/ref_dbs/Mitos2/Chordata/featureProt/{gene}.fas")), specifies the location on the MitoPilot docker image where the database is located, where {gene} is translated to individual gene names during processing. This default database location can be overridden by gene-specific paths added as named entires to the ref_db parameter, e.g. ref_db = list(default = "/ref_dbs/Mitos2/Chordata/featureProt/{gene}.fas", atp8 = "/ref_dbs/curation/atp8.fas"). However, it is important to note that these paths are relative to the execution environment and must be accessable from that environment. Building a custom docker image from the base MitoPilot image, includeding any additional reference databases within its file system, is one way to achieve this.

Curation Parameters

Currently, curation parameters are specified at project initialization and are applied to all samples within the project. By default, the parameters are set by the function, params_fish_mito(). Custimization of the parameters can be achieved by directly providing a complete named list, or by passing individual modifications to the params_fish_mito() function. For example, the default expected count of trnW genes could be increased to 2 and the default percent similarity of an “acceptable” PCG match could be reduced to 85 by initializing a new project with:

new_project(
  ...,
  curate_params = params_fish_mito(
    list(
      hit_threshold = 85,
      rules = list(
        trnW = list(
          count = 2
        )
      )
    )
  )
)

The full set of default curation parameters can be viewed in the MitoPilot GUI for each sample under “Curation Opts”, and is presented. At the moments, these values can not be edited for individual samples from within the GUI, but this feature will eventually be added. Developer Note: the curation parameters are stored in the project sqlite database as a base64 encoded text that must be parsed for reading / writing.

Validation

Following the automated curation process, samples are processed through a validation script, validate_fish_mito() by default. This script looks for anomolies in the final annotations relative to the expectations specified in the curation parameters and add notations to assist with manual curation using the MitoPilot GUI.

possible duplicate: The annotation occurs in the assembly more often than expected. All identical annotations will receive this flag and the user must determin which (if any) should be removed.
exceeds max overlap: The annotation overlaps one or more neighboring annotations on the same strand by a percentage of its length greater than the value set by the max_operlap parameter of the curation params. This represents a global maximum overlap threshold.
exceeds max start overlap: The start position of a PCG annotation overlaps one or more neighboring annotations on the same strand more than expected.
exceeds max stop overlap: The stop position of a PCG annotation overlaps one or more neighboring annotations on the same strand more than expected.
exceeds max length: The annotation length exceed the maximum expectation set in the curation params.
below min length: The annotation length is below the mimum expectation set in the curation params.
low coverage region: More than 5% of the assembly region defined by an annotation has less then 10x coverage.
high error region: More than 5% of the assembly region defined by an annotation has a raw read error rate greater than 5% (ie, more than 5% of raw reads than align to the base show disagreement on the base call).
internal stop codon: An internal stop codon was detected in a PCG annotation.
non-standard stop codon: A PCG stop codon other than those specified in the curation params was detected.
non-standard start codon: A PCG start codon other than those specified in the curation params was detected.
low reference similarity: The PCG curation reference database does not include any references with compositional similarity equal to ro greater than the hit_threshold set in the curation parameters (default = 90).
check reference start alignment: The PCG start position does not align exactly with the majority of the top hits from the reference database.
check reference stop alignment: The PCG stop position does not align exactly with the majority of the top hits from the reference database.