What tools are available for downstream analysis of 10x scATAC-seq in entirely new organisms?

Question

[Chilly]

Hi,

I'm working on a brand new animal. The animal has never even been published. We only have its genome file in the fasta format of this organism (containing the sequences of multiple contigs), and the gene file in gff3/gtf format (containing information such as transcript, exon, CDS, etc.).

I tried ArchR, Signac, SnapATAC, and scATAC-pro tools for downstream analysis of data generated by 10x cellranger-atac. But these tools are very unfriendly to the application of new species, and it is almost impossible to create the objects required by each tool with custom fasta and gff files.

Does anyone know of anyone who has published a non-model organism application of 10x scATAC-seq? And are there any downstream analysis tools that can customize species to recommend?

Thanks a lot!

i.e.: in Signac tutorial,

  counts <- Read10X_h5(filename = "../vignette_data/atac_v1_pbmc_10k_filtered_peak_bc_matrix.h5")
  metadata <- read.csv(
  file = "../vignette_data/atac_v1_pbmc_10k_singlecell.csv",
  header = TRUE,
  row.names = 1
  )

  chrom_assay <- CreateChromatinAssay(
  counts = counts,
  sep = c(":", "-"),
  genome = 'hg19',
  fragments = '../vignette_data/atac_v1_pbmc_10k_fragments.tsv.gz',
  min.cells = 10,
  min.features = 200,
  annotation = NULL
  )

  pbmc <- CreateSeuratObject(
  counts = chrom_assay,
  assay = "peaks",
  meta.data = metadata
  )

When I do the CreateChromatinAssay(), I could not directly use 'hg19' or anything else for 'genome =' argument. So I established a Seqinfo object containing basic information about the genome used:

sca.contig.size<-read.csv(file = '.../fasta/sca.contig.size.csv', header = F)   

Above, I used genome.fa.fai file in the cellranger reference file I created before.

Sca <- Seqinfo(seqnames=c(sca.contig.size$V1),
      seqlengths=c(sca.contig.size$V2),
      isCircular=c(rep(F, nrow(sca.contig.size))),
      genome="Sca")

And then I have to establish a 'GRanges' object or a set of 'GRanges' containing annotations for the genome used. This step is where I can't do anything about it. I have gff3 file but I don't know how to convert it to 'GRanges' object. The format introduction of 'GRanges' is very complicated, so I can only use other tools for object creation instead of manual editing like the Seqinfo object. So I tried:

if(!require(AcidGenomes))remotes::install_github("acidgenomics/AcidGenomes")

makeGRangesFromGFF(file=".../s.ca.primary.gff3", level = c("genes"))

Desperately,

Currently makeGRangesFromGFF() supports genomes from these sources:

Ensembl (GTF, GFF3).

GENCODE (GTF, GFF3).

RefSeq (GTF, GFF3).

FlyBase (GTF).

WormBase (GTF).

This goes back to the 'must use a published classic model' issue.

[mjsteinbaugh]

Hi @Chilly , I'm the developer the AcidGenomes package. Happy to help add support for the s.ca.primary.gff3 file you mentioned. Can you provide a copy of the file?

Best, Mike

Answer 1

[rpolicastro]

and it is almost impossible to create the objects required by each tool with custom fasta and gff files

Assuming that your fasta and gtf are not malformed, Signac should have worked for you since it uses fairly generic object types when passing genome annotation information to its functions. What problem did you have exactly with Signac?

[Chilly]

Hi rpo, Thanks for your reply and help. I've updated the problem description with more details to describe the problem I'm having. Regards, Di

[rpolicastro]

You should be able to load your GTF/GFF file into a GRanges object using import from the rtracklayer library.

[Chilly]

Thx, it worked.

[hjsunbio]

Hi di, may I know How you process the sc ATAC-seq on non-model organism, especially CreateChromatinAssay function. What should the GTF file like?