title: “Introduction of ‘geneHapR’” author: “Zhang RenLiang” date: “2023-04-05” output: rmarkdown::html_vignette vignette: > % % % editor_options: markdown: wrap: sentence
geneHapR is designed for gene haplotype statistics,
phenotype association and visualization.
Dataset required for haplotype statistic, visualization and phenotype association and the import function were listed in Table 1.
The genotype dataset is essential for haplotype identification and could be supplied in VCF, FASTA, P.link, HAPMAP and table format. The annotation were used for variants filtration and prepare schematic diagram.
Detailed information of individuals include phenotype data, group/category information and geo-coordinates. The phenotype data was used for comparison between different haplotypes. The group /category information was used for pie plot with haplotype network (eg. the second column in Table 4). And the geo-coordinates only used for demonstration of geographical distribution and include two columns: longitude and latitude (eg. the third and fourth column in Table 4).
Table 1: The required format of dataset and import functions for geneHapR
|Dataset||File format||Import function|
|VCF: *.vcf, *.vcf.gz;
Sequences: *.fa, .fasta;
p.link: (*.ped & *.map);
table (eg. Table2): .txt, *.csv
|GFF: .gff, .gff3,
BED4/BED6 (eg. Table3): *.bed
|table (eg. Table4): .txt, .csv||import_AccINFO()|
Table 2 is an example of genotypic data in table format: The first five column are fixed as chromosome name (CHROM), position (POS), reference nucleotide (REF), alter nucleotide (ALT) and additional information (INFO). Accession genotype should be in followed columns. “-” will be treated as Indel. “.” and “N” will be treated as missing data. Field in additional information column should be in format “tag=value”, and separated by semicolon “;”. Heterozygote should be looks like “A/G” or “A|G”.
Table 2: Table format of the genotypic dataset
Table 3 is an example of annotation file in BED6 format. As described at UCSC, the BED6 file contains 6 columns: 1) chromosome name, 2) chromosome start, 3) chromosome end, 4) name, 5) score and 6) strand. The BED4 contains the first 4 column of BED6.
BE NOTE THAT: the fourth column was used to define the name and types, which were separated by a space. For example, the first line of Table 3 indicates that: the genomic interval from 9154280 (exclude) to 9154821 (include) on Chr7 chromosome is CDS of “LOC_Os07g15770.1” and the strand is “negative”.
Table 3: An annotation example in BED6 format
|# CHROM||START||END||GENEID TYPE||.||STRAND|
Note: the red dot in fourth column indicate a space.
Table 4 is an example of detailed information of individuals, includes group/category, geo-coordinates and phenotype data. First column are names of accessions/individuals, phenotypic information are listed in followed columns.
Table 4: An example of accession detailed information dataset
The main results are
hapSummary class in R, consist of a matrix which could be
divided into three parts as shown in Fig.1, and some additional
Part I consists of only one column. And the first four lines were
fixed as CHROM (chromosome name), POS (position), INFO (additional
information) and ALLELE (allele). And followed lines are names of each
haplotype. Part II consists of at least one column, contains site
information (first four lines) and genotypes (followed lines). The part
hapResult consists of one column named as Accession,
hapSummary consists of two columns named as Accession
and freq (frequency of each haplotype).
The differences between
hapSummary is that each line of
indicate an accession/individual, and each line in
hapSummary indicate a haplotype.
geneHapR is schemed to submit to CRAN. If accepted, this
package could be installed with
geneHapR has not
published yet, if you use
geneHapR in your study, please
contact Zhang RenLiang
(Maintainer) (email: email@example.com) or Jia GuanQing
The first step is library the
geneHapR packages. I will
use the test data inside this package as an example for how to perform
statistics of a gene/range, visualization and phenotype association
There are two options to conduct a gene haplotype analysis starts from a VCF file or DNA sequences file. Thus a VCF file or DNA sequences file is necessary. However, the GFF, phenos and accession groups are strongly recommend for visualization and phenotype associations.
The import functions takes file path as input.
import_vcf() could import VCF file with surfix of “.vcf”
import_gff() import file format default as
import_seqs() file format default as “fasta”.
# import vcf file vcf <- import_vcf("your_vcf_file_path.vcf") # import gziped vcf file vcf <- import_vcf("your_vcf_file_path.vcf.gz")
# import GFFs gff <- import_gff("your_gff_file_path.gff", format = "GFF")
# import GFFs bed <- import_bed("your_gff_file_path.bed")
# import DNA sequences in fasta format seqs <- import_seqs("your_DNA_seq_file_path.fa", format = "fasta")
# import phynotype data pheno <- import_AccINFO("your_pheno_file_path.txt") pheno
## GrainWeight.2021 GrainWeight.2022 ## C1 16.76 18.76 ## C2 6.66 8.66 ## C3 7.80 16.30 ## C4 19.73 23.73 ## C5 11.95 16.95 ## C6 12.43 30.45
# import accession group/location information AccINFO <- import_AccINFO("accession_group_file_path.txt")
## Sensitive Type latitude longitude ## C1 Sensitive Mordern cultivar 65.216 33.677 ## C2 Resistance Mordern cultivar 65.216 33.677 ## C3 Mid Mordern cultivar 89.941 24.218 ## C4 Sensitive Mordern cultivar 89.941 24.218 ## C5 Resistance Mordern cultivar 89.941 24.218 ## C6 Mid Mordern cultivar 25.231 42.761
Be aware that the phenotype and accession group are effectively
tables. There are more than one ways to import a table format file with
Be Note that: a. the accession/individual names
located in first column; b. the first row contents
phenotype/accession_group names; c.
NA is allowed, it’s not
a wise option to replace
# import pheno from space ' ' delimed table pheno <- read.table("your_pheno_file_path.csv", header = TRUE, row.names = 1, comment.char = "#") # import pheno from ',' delimed table pheno <- read.csv("your_pheno_file_path.csv", header = TRUE, comment.char = "#")
There is a little work need to be done before haplotype calculations: (1) VCF filtration and (2) DNA sequences alignment.
There are three modes to filter a
vcfR object after
import VCF into ‘R’: a. by position; b. by annotation; c. by both of
# filter VCF by position vcf_f1 <- filter_vcf(vcf, mode = "POS", Chr = "scaffold_1", start = 4300, end = 5890) # filter VCF by annotation vcf_f2 <- filter_vcf(vcf, mode = "type", gff = gff, type = "CDS") # filter VCF by position and annotation vcf_f3 <- filter_vcf(vcf, mode = "both", Chr = "scaffold_1", start = 4300, end = 5890, gff = gff, type = "CDS")
It’s a time consuming work to import and manipulate a very large file
with ‘R’ on personal computer. It’ll be more efficiency to extract the
target ranges from origin VCF with
import. If your VCF file is just a few ‘MB’, this step was not necessary
Note: if extract more than one ranges, length of
output file names (
VCFout) must be equal with
# new VCF file will be saved to disk # extract a single gene/range from a large vcf filterLargeVCF(VCFin = "Ori.vcf.gz", VCFout = "filtered.vcf.gz", Chr = "scaffold_8", POS = c(19802,24501), override = TRUE) # extract multi genes/ranges from large vcf filterLargeVCF(VCFin = "Ori.vcf.gz", # surfix should be .vcf.gz or .vcf VCFout = c("filtered1.vcf.gz", # surfix should be .vcf.gz or .vcf "filtered2.vcf.gz", "filtered3.vcf.gz"), Chr = c("scaffold_8", "scaffold_8", "scaffold_7"), POS = list(c(19802,24501), c(27341,28949), c(38469,40344)), override = TRUE) # if TRUE, existed file will be override without warning
The origin DNA sequences must be aligned and trimmed due to haplotype
calculation need all sequences have same length. Those operations could
be done with
geneHapR. I still suggest users align and trim
DNA sequences with Mega software and then save the
result as FASTA format before import them into ‘R’.
# sequences alignment seqs <- allignSeqs(seqs, quiet = TRUE) # sequences trim seqs <- trimSeqs(seqs, minFlankFraction = 0.1) seqs
hap <- filter_hap(hapSummary, rm.mode = c("position", "accession", "haplotype", "freq"), position.rm = c(4879, 4950), accession.rm = c("C1", "C9"), haplotype.rm = c("H009", "H008"), freq.min = 5)
As mentioned before, haplotype could be calculated from VCF or
genotype of most sites should be known and homozygous, still, a few site
are unknown or heterozygous due to chromosome variant or error cased by
sequencing or SNP calling or gaps or other reasons. It’s a hard decision
whether to drop accessions/individuals contains heterozygous or unknown
sites for every haplotype analysis. Hence, I leave the choice to
Calculate haplotype result from VCF.
hapResult <- vcf2hap(vcf, hapPrefix = "H", hetero_remove = TRUE, na_drop = TRUE) hapResult
## Accessions: ## All: 37 ## Removed: 4 C12, C16, C29, C33 ## Remain: 33 ## ## hap frequences: ## H001 H002 H003 H004 H005 H006 H007 H008 H009 ## 10 8 4 4 2 2 1 1 1 ## ## Options: ## hapPrefix: H ## CHROM: scaffold_1 ## POS: 4300-6856 ## hetero_remove: YES ## NA_remove: YES ## ## # A tibble: 37 × 11 ## Hap `4300` `6856` `5209` `5213` `4910` `4879` `4345` `4950` `5037` Accession ## <chr> <chr> <chr> <chr> <chr> <chr> <chr> <chr> <chr> <chr> <chr> ## 1 CHR scaff… scaff… scaff… scaff… scaff… scaff… scaff… scaff… scaff… "" ## 2 POS 4300 6856 5209 5213 4910 4879 4345 4950 5037 "" ## 3 INFO <NA> <NA> <NA> <NA> <NA> <NA> <NA> <NA> <NA> "" ## 4 ALLE… G/C A/G A/AC C/G GCCTA… T/A,G T/A,GG T/AA,… A/AA,… "" ## 5 H001 G A A C GCCTA T T T A "C8" ## 6 H001 G A A C GCCTA T T T A "C9" ## 7 H001 G A A C GCCTA T T T A "C11"
Calculate haplotype result from aligned DNA sequences.
hapResult <- seqs2hap(seqs, Ref = names(seqs), hapPrefix = "H", hetero_remove = TRUE, na_drop = TRUE, maxGapsPerSeq = 0.25)
Before visualization, there were a few details need to be adjusted. eg. add annotations and adjust position of “ATG”
hapResult was calculated from
object, the INFO was taken from
field. The VCF INFO may missing some annotations. or contents
format was inappropriate to display. Further more, INFO
contents nothing if
hapResult was generated from sequences.
Here, we can introduce/replace the origin INFO by
Note that: length of
values must be
equal with number of sites.
Let’s see how mant sites contains in the
# Chech number of sites conclude in hapResult sites(hapResult)
##  9
Now we replace the old INFO field with new tag named as “PrChange”.
# add annotations to INFO field hapResult <- addINFO(hapResult, tag = "PrChange", values = rep(c("C->D", "V->R", "G->N"),3), replace = TRUE)
Here, we add a tag named as “CDSChange” followed the old INFO.
# To replace the origin INFO by set 'replace' as TRUE hapResult <- addINFO(hapResult, tag = "CDSChange", values = rep(c("C->A", "T->C", "G->T"),3), replace = FALSE)
This function was only used to adjust the position of “ATG” to 0 and hence convert the gene on negative strand to positive strand.
Be note that: GFF and hapResult need to adjust position of ATG with the same parameters.
# set ATG position as zero in gff newgff <- gffSetATGas0(gff = gff, hap = hapResult, geneID = "test1G0387", Chr = "scaffold_1", POS = c(4300, 7910)) # set position of ATG as zero in hapResult/hapSummary newhap <- hapSetATGas0(gff = gff, hap = hapResult, geneID = "test1G0387", Chr = "scaffold_1", POS = c(4300, 7910))
hapResultsummary and visualization
Once we have the
hapResult object, can we summarize and
hapResult by interact with annotations and
Now, we have the
hapResult object with INFOs we want
display in next step. The
hap_summary() function convert
the object of
hapResult class, which is a long table
format, into a short table belong to
hapSummary class. In
hapResult each row represent a accession, while each row
represents a hap in
hapSummary <- hap_summary(hapResult) hapSummary
## ## Accssions: 33 ## Sites: 9 ## Indels: 5 ## SNPs: 4 ## ## Haplotypes: 9 ## H001 10 C8, C9, C11, C14, C18, C25, ... ## H002 8 C5, C15, C17, C19, C22, C32, ... ## H003 4 C6, C20, C23, C37 ## H004 4 C7, C13, C24, C30 ## H005 2 C4, C21 ## H006 2 C10, C27 ## H007 1 C1 ## H008 1 C2 ## H009 1 C3 ## ## Options: ## hapPrefix: H ## CHROM: scaffold_1 ## POS: 4300-6856 ## hetero_remove: YES ## NA_remove: YES ## ## # A tibble: 13 × 12 ## Hap `4300` `6856` `5209` `5213` `4910` `4879` `4345` `4950` `5037` Accession ## <chr> <chr> <chr> <chr> <chr> <chr> <chr> <chr> <chr> <chr> <chr> ## 1 CHR scaff… scaff… scaff… scaff… scaff… scaff… scaff… scaff… scaff… "Haploty… ## 2 POS 4300 6856 5209 5213 4910 4879 4345 4950 5037 "Individ… ## 3 INFO PrCha… PrCha… PrCha… PrCha… PrCha… PrCha… PrCha… PrCha… PrCha… "Variant… ## 4 ALLE… G/C A/G A/AC C/G GCCTA… T/A,G T/A,GG T/AA,… A/AA,… "" ## 5 H001 G A A C GCCTA T T T A "C8;C9;C… ## 6 H002 C A A G A T T T A "C5;C15;… ## 7 H003 G A AC G A A A AA AA "C6;C20;…
Let’s see how to visualization of our haplotype results.
At first let’s display the
hapSummary as a table. In
this table like figure we can see all the variants and their positions,
haplotypes and their frequencies.
Also we can add an annotation, “CDSChange”, to the table by assign
INFO_tag. It’s your responsibility to verify whether
the INFO_tag was existed in the INFO field.
# add one annotation plotHapTable(hapSummary, hapPrefix = "H", INFO_tag = "CDSChange", tag_name = "CDS", displayIndelSize = 1, angle = 45, replaceMultiAllele = TRUE, ALLELE.color = "grey90")
Now let’s add another
INFO_tag named as “PrChange”.
# add multi annotation plotHapTable(hapSummary, hapPrefix = "H", INFO_tag = c("CDSChange", "PrChange"), displayIndelSize = 1, angle = 45, replaceMultiAllele = TRUE, ALLELE.color = "grey90")
tag_name was used to replace the character if
INFO_tag was too long.
# add multi annotation plotHapTable(hapSummary, hapPrefix = "H", INFO_tag = c("CDSChange", "PrChange"), tag_name = c("CDS", "Pr"), displayIndelSize = 1, angle = 45, replaceMultiAllele = TRUE, ALLELE.color = "grey90")
I think it’s a good idea to figure out where are the variants by marking them on gene model.
displayVarOnGeneModel(hapSummary, gff, Chr = "scaffold_1", startPOS = 4300, endPOS = 7910, type = "pin", cex = 0.7, CDS_h = 0.05, fiveUTR_h = 0.02, threeUTR_h = 0.01)
hapNetcalculation and visualization
hapNet could be generated from object of
hapSummary class. The accession group information could be
attached in this step.
hapNet <- get_hapNet(hapSummary, AccINFO = AccINFO, groupName = "Type")
Once we have the
hapNet object, we can plot it with
# plot haploNet plotHapNet(hapNet, size = "freq", # circle size scale = "log2", # scale circle with 'log10(size + 1)' cex = 0.8, # size of hap symbol col.link = 2, # link colors link.width = 2, # link widths show.mutation = 2, # mutation types one of c(0,1,2,3) legend = c(-12.5, 7)) # legend position
Now we get the haplotype result. There is a new question emerged: how did those main haplotypes distributed, are they related to geography?
# library(mapdata) # library(maptools) hapDistribution(hapResult, AccINFO = AccINFO, LON.col = "longitude", LAT.col = "latitude", hapNames = c("H001", "H002", "H003"), legend = TRUE)
Finally, let’s see which haplotype has superiority at particular area by interact with phynotype.
Here are two options, merged or separated, to organized the heatmap
of p-values and violin plot. The figure as an object of
ggplot2, which means user could add/modified figure
Here is an example for merged arrangement:
results <-hapVsPheno(hapResult, hapPrefix = "H", title = "This is title", mergeFigs = TRUE, pheno = pheno, phenoName = "GrainWeight.2021", minAcc = 3) plot(results$figs)
An example for separated plot:
results <- hapVsPheno(hap = hapResult, hapPrefix = "H", title = "This is title", pheno = pheno, phenoName = "GrainWeight.2021", minAcc = 3, mergeFigs = FALSE) plot(results$fig_pvalue)