normalize: Normalize calls

In sandmanns/appreci8R: appreci8R: an R/Bioconductor package for filtering SNVs and short indels with high sensitivity and high PPV

Description

appreci8R combines and filters the output of different variant calling tools according to the 'appreci8'-algorithm. In the 2nd analysis step, all calls are normalized with respect to reporting indels, MNVs, reporting of several alternate alleles and reporting of complex indels. A GRanges object with all normalized calls is returned.

Usage

normalize(output_folder, caller_name, target_calls, caller_indels_pm,
          caller_mnvs)

Arguments

output_folder	The folder to write the output files into. If an empty string is provided, no files are written out.
caller_name	Name of the variant calling tool (only necessary if an output folder is provided).
target_calls	List of data.frames. One list element per sample. FilterTarget()-output can directly be taken as input.
caller_indels_pm	Deletions are reported with a “minus” (e.g. C > -A), insertions are reported with a “plus” (e.g. C > +A) (TRUE or FALSE).
caller_mnvs	MNVs are reported (e.g. CA > GT instead of C > G and A > T; or CGAG > TGAT instead of C > T and G > T) (TRUE or FALSE).

Details

The function normalize covers two to four normalization steps:

1) Check alternative bases: Calls containing a “comma” are split up. A call like C > A,G is converted to C > A and an additional C > G call. This enables evaluation of the output of different callers while caller1 reports C > A,G, caller2 only reports C > A and caller3 only reports C > G. This normalization step is always performed.

2) Find string differences: Calls are checked for un-mutated bases. The smallest option of reporting a variant at the left-most position is chosen. For example, CAAAC > CAAC is converted to CA > C. This normalization step is always performed.

3) Convert indels: If deletions are reported with a “minus” and insertions are reported with a “plus”, these are converted. An deletion like C > -G is converted to CG > C, while an insertion like C > +G is converted to C > CG. This normalization step is only performed if caller_indels_pm is TRUE.

4) Convert MNVs: If MNVs are reported, these are converted. This enables evaluation of the output of different callers if not all callers report all mutations being part of an MNV. A call like CA > GT is split up to a C > G and an A > T variant. But also a call like CGAG > TGAT is split up to C > T and G > T (G > G and A > A are not reported as they do not pass the normalization step “Find string differences”). This normalization step is only performed if caller_mnvs is TRUE.

Value

A GRanges object is returned (metadata columns: SampleID, Ref, Alt).

If an output folder is provided, the output is saved as <caller_name>.normalized.txt.

Author(s)

Sarah Sandmann <[e-mail]>

See Also

appreci8R, appreci8Rshiny, filterTarget, annotate, combineOutput, evaluateCovAndBQ, determineCharacteristics, finalFiltration

Examples

sample1<-data.frame(SampleID = c("Sample1","Sample1","Sample1"),
                    Chr = c("2","17","X"),
                    Pos = c(25469502,7579472,15838366),
                    Ref = c("CAG","G","C"),
                    Alt = c("TAT","C","T,A"))
sample2<-data.frame(SampleID = c("Sample2","Sample2","Sample2","sample2"),
                    Chr = c("4","12","12","21"),
                    Pos = c(106196951,12046289,12046341,36164405),
                    Ref = c("A","C","A","GGG"),
                    Alt = c("G","+AAAG","G","TGG"))
input<-list(sample1, sample2)

normalized<-normalize("", "", input, TRUE, TRUE)

sandmanns/appreci8R documentation built on Dec. 7, 2020, 12:32 a.m.