filterMT: Filter SummarizedExperiment or DataFrame values based on its...
In MTseeker: Bioconductor Tools for Human Mitochondrial Variant Analysis

Description Usage Arguments Details Value Examples

Griffin et al. (Genetics in Medicine 2014) recommends 20x coverage for mtDNA sequencing to have comparable error rates to Sanger sequencing. By default, that is the cutoff applied here to ensure halfway decent variant annotation.

1	filterMT(DFSE, minCovg = 20, fpFilter = FALSE, NuMT = FALSE)

`DFSE`	a DataFrame/SummarizedExperiment with colData()$'mtCovg'
`minCovg`	minimum covg (20, cf. Griffin, Genetics in Medicine 2014)
`fpFilter`	apply Triska's homopolymer false positive filter? (FALSE)
`NuMT`	apply the 0.03 VAF NuMT filter from Ju (GR 2015)? (FALSE)

Triska (Cancer Res, in revision) suggests a small number of masked regions where homopolymers can be a problem; these are avoided if fpFilter

The NuMT filtration step (Ju, in eLife 2014, suggests a variant allele cutoff of 0.03 to avoid false positive calls from nuclear-mitochondrial translocated or 'NuMT' fragments) is also a useful tool to cut down on nonsensical calls, although it may be important to use caution as low heteroplasmy can also resolve into apparent near-homoplasmy at the single-cell level, at least in our (TJT & co) experience.

As a consequence of the Wild West nature for published methods of high- throughput mitochondrial sequence variant analysis at the time of writing (2018), the default for this function is to filter on coverage only, as the user is expected to determine what additional filters to apply. We could envision changing these defaults down the road as standards congeal.

If DFSE is an MVRanges[List], the function will call filterMTvars instead.

a filtered SE or data.frame