Support for GRCh38?

[MikeHala]

Dear Vincent,

Thank you for developing and maintaining ExomeDepth!

Are there any plans to support GRCh38, more specifically have you crated by any chance
exons.GRCh38 Positions of exons on build GRCh38 of the human genome
exons.GRCh38 Positions of exons on build GRCh38 of the human genome and on chromosome X
genes.GRCh38 Positions of genes on build GRCh38 of the human genome

Best,
Mike

[Maroua2906]

Hi Mike,
Have you tested exomedepth using hg38 data?
Thank you

[skgs1970]

I have recently used the GCA_000001405.15_GRCh38_full_analysis_set.refseq_annotation.gff.gz file and converted it into a bed file like exons.hg19. It works well. But we are still testing it and comparing it with the GATK CNV caller. The size of the database increases a bit. But it works.

[CODDLECODD]

Hi,
I prefer to use the bed target file of what you enrich for, then you don't miss anything. For example we use Twist for exomes so we use their bed-files:
twist.hg38 <- read.csv(twist_hg38.bed) with the columns chromosome,start,end,name
then: my.counts <- getBamCounts(bed.frame = twist.hg38

Similarly you can download both hg38 genes and exon bed-files from UCSC Table (genome.ucsc.edu/cgi-bin/hgTables), variant files from dgv or gnomAD SV instead of Conrad (dgv.tcag.ca/dgv/app/downloads) and use for annotation.

We use ExomeDepth in the clinic and it outperforms CoNIFER which we also use, and is equally good as CNVpytor on WGS. It would be interesting to hear about the GATK CNV caller comparison. The best performance we see is when ExomeDepth is used on a small cancer panel (Twist custom design) where we cover the entire gene regions, including introns, and we divide the target into a 200bp sized bed frame to increase resolution.

PS. The largest annotation file, exons.hg38.bed, is 92Mb.

Cheers
Joakim Klar