phyloseq_nmds: phyloseq_nmds.R annotate

annotate phyloseq_nmds.R @ 2:20adf95eb758 draft

Updated phyloseq_nmds.xml to allow for new datatype.

author	simon-gladman
date	Fri, 20 Jul 2018 00:23:53 -0400
parents	e376a618eb9f
children

rev	line source
0 b4606394e7ec planemo upload simon-gladman parents: diff changeset	1 library('getopt')
b4606394e7ec planemo upload simon-gladman parents: diff changeset	2 library('ape')
b4606394e7ec planemo upload simon-gladman parents: diff changeset	3 suppressPackageStartupMessages(library('phyloseq'))
1 e376a618eb9f Updated phyloseq_nmds.R to allow for non standard BIOM files. simon-gladman parents: 0 diff changeset	4 library(biomformat)
e376a618eb9f Updated phyloseq_nmds.R to allow for non standard BIOM files. simon-gladman parents: 0 diff changeset	5 library(plyr)
0 b4606394e7ec planemo upload simon-gladman parents: diff changeset	6 Sys.setenv("DISPLAY"=":1")
b4606394e7ec planemo upload simon-gladman parents: diff changeset	7 library("ggplot2")
2 20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	8 suppressPackageStartupMessages(library("doParallel"))
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	9 ncores = ceiling(detectCores() * 0.8)
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	10 registerDoParallel(cores=ncores)
0 b4606394e7ec planemo upload simon-gladman parents: diff changeset	11
b4606394e7ec planemo upload simon-gladman parents: diff changeset	12 options(warn=-1)
b4606394e7ec planemo upload simon-gladman parents: diff changeset	13
b4606394e7ec planemo upload simon-gladman parents: diff changeset	14 theme_set(theme_bw())
b4606394e7ec planemo upload simon-gladman parents: diff changeset	15
b4606394e7ec planemo upload simon-gladman parents: diff changeset	16 #http://saml.rilspace.com/creating-a-galaxy-tool-for-r-scripts-that-output-images-and-pdfs
b4606394e7ec planemo upload simon-gladman parents: diff changeset	17 #http://joey711.github.io/phyloseq-demo/phyloseq-demo.html
b4606394e7ec planemo upload simon-gladman parents: diff changeset	18 option_specification = matrix(c(
b4606394e7ec planemo upload simon-gladman parents: diff changeset	19 'otu_table','o',2,'character',
b4606394e7ec planemo upload simon-gladman parents: diff changeset	20 'tax_table','t',2,'character',
2 20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	21 'meta_table','m',2,'character',
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	22 'biom','b',2,'character',
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	23 'filter','f',2,'numeric',
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	24 'subset','s',2,'character',
0 b4606394e7ec planemo upload simon-gladman parents: diff changeset	25 'method','n',2,'character',
b4606394e7ec planemo upload simon-gladman parents: diff changeset	26 'distance','d',2,'character',
b4606394e7ec planemo upload simon-gladman parents: diff changeset	27 'kingdom','k',2,'character',
2 20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	28 'plottype','e',2,'numeric',
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	29 'cutoff','c',2,'numeric',
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	30 'category','g',2,'numeric',
0 b4606394e7ec planemo upload simon-gladman parents: diff changeset	31 'keep','p',2,'numeric',
b4606394e7ec planemo upload simon-gladman parents: diff changeset	32 'outdir','r',2,'character',
b4606394e7ec planemo upload simon-gladman parents: diff changeset	33 'htmlfile','h',2,'character'
b4606394e7ec planemo upload simon-gladman parents: diff changeset	34 ),byrow=TRUE,ncol=4);
b4606394e7ec planemo upload simon-gladman parents: diff changeset	35
b4606394e7ec planemo upload simon-gladman parents: diff changeset	36
b4606394e7ec planemo upload simon-gladman parents: diff changeset	37 options <- getopt(option_specification);
b4606394e7ec planemo upload simon-gladman parents: diff changeset	38 options(bitmapType="cairo")
b4606394e7ec planemo upload simon-gladman parents: diff changeset	39
b4606394e7ec planemo upload simon-gladman parents: diff changeset	40 if (!is.null(options$outdir)) {
b4606394e7ec planemo upload simon-gladman parents: diff changeset	41 # Create the directory
b4606394e7ec planemo upload simon-gladman parents: diff changeset	42 dir.create(options$outdir,FALSE)
b4606394e7ec planemo upload simon-gladman parents: diff changeset	43 }
b4606394e7ec planemo upload simon-gladman parents: diff changeset	44
b4606394e7ec planemo upload simon-gladman parents: diff changeset	45
b4606394e7ec planemo upload simon-gladman parents: diff changeset	46
b4606394e7ec planemo upload simon-gladman parents: diff changeset	47 method<-options$method
b4606394e7ec planemo upload simon-gladman parents: diff changeset	48 cutoff_value<-options$cutoff
b4606394e7ec planemo upload simon-gladman parents: diff changeset	49 ### select a kingdom for phyloseq plot (e.g., "phylum")
b4606394e7ec planemo upload simon-gladman parents: diff changeset	50 #kingdom_str<-colnames(tax_table)[options$kingdom]
b4606394e7ec planemo upload simon-gladman parents: diff changeset	51 kingdom_str<-options$kingdom
b4606394e7ec planemo upload simon-gladman parents: diff changeset	52 distance<-options$distance
b4606394e7ec planemo upload simon-gladman parents: diff changeset	53 keep<-options$keep
2 20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	54 plottype<-options$plottype
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	55 filter<-options$filter
0 b4606394e7ec planemo upload simon-gladman parents: diff changeset	56
2 20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	57 ### prepare the directory and file name
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	58 pdffile <- gsub("[ ]+", "", paste(options$outdir,"/pdffile.pdf"))
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	59 pngfile_nmds <- gsub("[ ]+", "", paste(options$outdir,"/nmds.png"))
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	60 pngfile_nmds_facet <- gsub("[ ]+", "", paste(options$outdir,"/nmds_facet.png"))
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	61 htmlfile <- gsub("[ ]+", "", paste(options$htmlfile))
0 b4606394e7ec planemo upload simon-gladman parents: diff changeset	62
1 e376a618eb9f Updated phyloseq_nmds.R to allow for non standard BIOM files. simon-gladman parents: 0 diff changeset	63
e376a618eb9f Updated phyloseq_nmds.R to allow for non standard BIOM files. simon-gladman parents: 0 diff changeset	64 ### This function accepts different two different type of BIOM file format
e376a618eb9f Updated phyloseq_nmds.R to allow for non standard BIOM files. simon-gladman parents: 0 diff changeset	65 readBIOM<-function(inBiom){
e376a618eb9f Updated phyloseq_nmds.R to allow for non standard BIOM files. simon-gladman parents: 0 diff changeset	66 tryCatch({
2 20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	67 phyloseq_obj<-import_biom(inBiom,parallel=TRUE)
1 e376a618eb9f Updated phyloseq_nmds.R to allow for non standard BIOM files. simon-gladman parents: 0 diff changeset	68 return(phyloseq_obj)
e376a618eb9f Updated phyloseq_nmds.R to allow for non standard BIOM files. simon-gladman parents: 0 diff changeset	69 },
e376a618eb9f Updated phyloseq_nmds.R to allow for non standard BIOM files. simon-gladman parents: 0 diff changeset	70 error=function(e){
e376a618eb9f Updated phyloseq_nmds.R to allow for non standard BIOM files. simon-gladman parents: 0 diff changeset	71 biom_obj<-read_biom(inBiom)
e376a618eb9f Updated phyloseq_nmds.R to allow for non standard BIOM files. simon-gladman parents: 0 diff changeset	72
e376a618eb9f Updated phyloseq_nmds.R to allow for non standard BIOM files. simon-gladman parents: 0 diff changeset	73 otu_matrix = as(biom_data(biom_obj), "matrix")
e376a618eb9f Updated phyloseq_nmds.R to allow for non standard BIOM files. simon-gladman parents: 0 diff changeset	74 OTU_TABLE = otu_table(otu_matrix, taxa_are_rows=TRUE)
e376a618eb9f Updated phyloseq_nmds.R to allow for non standard BIOM files. simon-gladman parents: 0 diff changeset	75
e376a618eb9f Updated phyloseq_nmds.R to allow for non standard BIOM files. simon-gladman parents: 0 diff changeset	76 taxonomy_matrix = as.matrix(observation_metadata(biom_obj), rownames.force=TRUE)
e376a618eb9f Updated phyloseq_nmds.R to allow for non standard BIOM files. simon-gladman parents: 0 diff changeset	77 TAXONOMY_TABLE = tax_table(taxonomy_matrix)
e376a618eb9f Updated phyloseq_nmds.R to allow for non standard BIOM files. simon-gladman parents: 0 diff changeset	78
e376a618eb9f Updated phyloseq_nmds.R to allow for non standard BIOM files. simon-gladman parents: 0 diff changeset	79 metadata.temp<-sample_metadata(biom_obj)
e376a618eb9f Updated phyloseq_nmds.R to allow for non standard BIOM files. simon-gladman parents: 0 diff changeset	80 METADATA_TABLE<-plyr::ldply(metadata.temp, rbind)
2 20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	81 rownames(METADATA_TABLE)<-as.character(METADATA_TABLE$.id)
1 e376a618eb9f Updated phyloseq_nmds.R to allow for non standard BIOM files. simon-gladman parents: 0 diff changeset	82
e376a618eb9f Updated phyloseq_nmds.R to allow for non standard BIOM files. simon-gladman parents: 0 diff changeset	83 phyloseq_obj = phyloseq(OTU_TABLE, TAXONOMY_TABLE,sample_data(METADATA_TABLE))
e376a618eb9f Updated phyloseq_nmds.R to allow for non standard BIOM files. simon-gladman parents: 0 diff changeset	84 return(phyloseq_obj)
e376a618eb9f Updated phyloseq_nmds.R to allow for non standard BIOM files. simon-gladman parents: 0 diff changeset	85 }
e376a618eb9f Updated phyloseq_nmds.R to allow for non standard BIOM files. simon-gladman parents: 0 diff changeset	86 )
e376a618eb9f Updated phyloseq_nmds.R to allow for non standard BIOM files. simon-gladman parents: 0 diff changeset	87 }
e376a618eb9f Updated phyloseq_nmds.R to allow for non standard BIOM files. simon-gladman parents: 0 diff changeset	88
e376a618eb9f Updated phyloseq_nmds.R to allow for non standard BIOM files. simon-gladman parents: 0 diff changeset	89
2 20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	90 create_OTU_PDF<-function(pdf_file,phyloseq_obj,phyloseq_ord,kingdom_str,htmlfile,pngfile_nmds,pngfile_nmds_facet){
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	91 pdf(pdf_file);
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	92 p1<-plot_ordination(physeq,physeq_ord,type="taxa",color="Phylum",title="taxa")
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	93 print(p1)
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	94 p2<-plot_ordination(physeq,physeq_ord,type="taxa",color="Phylum",title="taxa") + facet_wrap(formula(paste('~',kingdom_str)),3)
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	95 print(p2)
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	96 garbage<-dev.off();
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	97
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	98 #png('nmds.png')
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	99 bitmap(pngfile_nmds,"png16m")
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	100 p3<-plot_ordination(physeq,physeq_ord,type="taxa",color="Phylum",title="taxa")
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	101 print(p3)
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	102 garbage<-dev.off()
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	103
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	104 #png('nmds_facet.png')
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	105 bitmap(pngfile_nmds_facet,"png16m")
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	106 p4<-plot_ordination(physeq,physeq_ord,type="taxa",color="Phylum",title="taxa") + facet_wrap(formula(paste('~',kingdom_str)),3)
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	107 print(p4)
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	108 garbage<-dev.off()
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	109
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	110 create_HTML_1(htmlfile)
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	111 }
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	112
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	113 create_SAMPLE_PDF<-function(pdf_file,phyloseq_obj,phyloseq_ord,htmlfile,pngfile_nmds,category_type){
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	114 pdf(pdf_file);
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	115 p <- plot_ordination(physeq_filter, physeq_ord, type="samples", color=category_type)
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	116 p <- p + geom_point(aes(fill=category_type)) + geom_point(size=5) + ggtitle("samples")
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	117 print(p)
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	118 garbage<-dev.off();
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	119
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	120 #png('nmds.png')
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	121 bitmap(pngfile_nmds,"png16m")
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	122 p1 <- plot_ordination(physeq_filter, physeq_ord, type="samples", color=category_type)
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	123 p1 <- p1 + geom_point(aes(fill=category_type)) + geom_point(size=5) + ggtitle("samples")
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	124 print(p1)
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	125 garbage<-dev.off();
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	126
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	127 create_HTML_2(htmlfile)
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	128 }
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	129
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	130 create_HTML_1<-function(htmlfile){
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	131 htmlfile_handle <- file(htmlfile)
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	132 html_output = c('<html><body>',
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	133 '<table align="center>',
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	134 '<tr>',
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	135 '<td valign="middle" style="vertical-align:middle;">',
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	136 '<a href="pdffile.pdf"><img src="nmds.png"/></a>',
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	137 '</td>',
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	138 '</tr>',
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	139 '<tr>',
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	140 '<td valign="middle" style="vertical-align:middle;">',
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	141 '<a href="pdffile.pdf"><img src="nmds_facet.png"/></a>',
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	142 '</td>',
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	143 '</tr>',
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	144 '</table>',
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	145 '</html></body>');
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	146 writeLines(html_output, htmlfile_handle);
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	147 close(htmlfile_handle);
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	148 }
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	149
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	150 create_HTML_2<-function(htmlfile){
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	151 htmlfile_handle <- file(htmlfile)
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	152 html_output = c('<html><body>',
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	153 '<table align="center>',
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	154 '<tr>',
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	155 '<td valign="middle" style="vertical-align:middle;">',
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	156 '<a href="pdffile.pdf"><img src="nmds.png"/></a>',
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	157 '</td>',
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	158 '</tr>',
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	159 '</table>',
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	160 '</html></body>');
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	161 writeLines(html_output, htmlfile_handle);
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	162 close(htmlfile_handle);
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	163 }
1 e376a618eb9f Updated phyloseq_nmds.R to allow for non standard BIOM files. simon-gladman parents: 0 diff changeset	164
0 b4606394e7ec planemo upload simon-gladman parents: diff changeset	165 if(!is.null(options$biom)){
b4606394e7ec planemo upload simon-gladman parents: diff changeset	166
1 e376a618eb9f Updated phyloseq_nmds.R to allow for non standard BIOM files. simon-gladman parents: 0 diff changeset	167 #physeq<-import_biom(options$biom)
e376a618eb9f Updated phyloseq_nmds.R to allow for non standard BIOM files. simon-gladman parents: 0 diff changeset	168 physeq<-readBIOM(options$biom)
2 20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	169
0 b4606394e7ec planemo upload simon-gladman parents: diff changeset	170 if(length(rank_names(physeq)) == 8){
2 20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	171 tax_table(physeq) <- tax_table(physeq)[,-1]
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	172 colnames(tax_table(physeq)) <- c("Kingdom", "Phylum", "Class", "Order", "Family", "Genus", "Species")
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	173 } else {
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	174 colnames(tax_table(physeq)) <- c("Kingdom", "Phylum", "Class", "Order", "Family", "Genus", "Species")
0 b4606394e7ec planemo upload simon-gladman parents: diff changeset	175 }
2 20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	176
0 b4606394e7ec planemo upload simon-gladman parents: diff changeset	177 ### select column name from sample table for nmds plot
2 20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	178 ## which(colnames(sample_data(biom)) == "vegetation_type_id")
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	179 #category_type<-colnames(sample_data(physeq))[options$subset]
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	180 category_type <- options$subset
0 b4606394e7ec planemo upload simon-gladman parents: diff changeset	181
b4606394e7ec planemo upload simon-gladman parents: diff changeset	182 ### obtain the unique value in the selected column from sample table
b4606394e7ec planemo upload simon-gladman parents: diff changeset	183 category_option<-unique(sample_data(physeq))[,options$subset]
b4606394e7ec planemo upload simon-gladman parents: diff changeset	184
b4606394e7ec planemo upload simon-gladman parents: diff changeset	185 }else{
b4606394e7ec planemo upload simon-gladman parents: diff changeset	186
b4606394e7ec planemo upload simon-gladman parents: diff changeset	187 ### read the data into correct data type to create phyloseq object
b4606394e7ec planemo upload simon-gladman parents: diff changeset	188 otu_table<-as.matrix(read.table(options$otu_table,header=T,sep="\t"))
b4606394e7ec planemo upload simon-gladman parents: diff changeset	189 tax_table<-as.matrix(read.table(options$tax_table,header=T,sep="\t"))
b4606394e7ec planemo upload simon-gladman parents: diff changeset	190 sample_table<-read.table(options$meta_table,header=T,sep="\t",stringsAsFactors=F)
b4606394e7ec planemo upload simon-gladman parents: diff changeset	191
b4606394e7ec planemo upload simon-gladman parents: diff changeset	192
b4606394e7ec planemo upload simon-gladman parents: diff changeset	193 ### select column name from sample table for nmds plot
b4606394e7ec planemo upload simon-gladman parents: diff changeset	194 category_type<-colnames(sample_table)[options$category]
b4606394e7ec planemo upload simon-gladman parents: diff changeset	195
b4606394e7ec planemo upload simon-gladman parents: diff changeset	196 ### obtain the unique value in the selected column from sample table
b4606394e7ec planemo upload simon-gladman parents: diff changeset	197 category_option<-unique(sample_table[,options$category])
b4606394e7ec planemo upload simon-gladman parents: diff changeset	198
b4606394e7ec planemo upload simon-gladman parents: diff changeset	199
b4606394e7ec planemo upload simon-gladman parents: diff changeset	200 ### create a sample object for phyloseq
b4606394e7ec planemo upload simon-gladman parents: diff changeset	201 sample_object<-sample_data(sample_table)
b4606394e7ec planemo upload simon-gladman parents: diff changeset	202
b4606394e7ec planemo upload simon-gladman parents: diff changeset	203 ### create otu object for phyloseq
b4606394e7ec planemo upload simon-gladman parents: diff changeset	204 OTU<-otu_table(otu_table, taxa_are_rows = TRUE)
b4606394e7ec planemo upload simon-gladman parents: diff changeset	205
b4606394e7ec planemo upload simon-gladman parents: diff changeset	206 ### create tax object for phyloseq
b4606394e7ec planemo upload simon-gladman parents: diff changeset	207 TAX<-tax_table(tax_table)
b4606394e7ec planemo upload simon-gladman parents: diff changeset	208
b4606394e7ec planemo upload simon-gladman parents: diff changeset	209 ### create a phyloseq object
b4606394e7ec planemo upload simon-gladman parents: diff changeset	210 physeq = phyloseq(OTU,TAX,sample_object)
b4606394e7ec planemo upload simon-gladman parents: diff changeset	211 }
b4606394e7ec planemo upload simon-gladman parents: diff changeset	212 ### select a kingdom for phyloseq plot (e.g., "phylum")
b4606394e7ec planemo upload simon-gladman parents: diff changeset	213 #kingdom_str<-colnames(tax_table)[options$kingdom]
b4606394e7ec planemo upload simon-gladman parents: diff changeset	214 #kingdom_str<-options$kingdom
2 20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	215
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	216
0 b4606394e7ec planemo upload simon-gladman parents: diff changeset	217 ### filtering OTUs based on cutoff value (e.g., 5)
2 20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	218 #physeq_temp =genefilter_sample(physeq, filterfun_sample(function(x) x > cutoff_value), A=0.1*nsamples(physeq))
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	219 physeq_temp =genefilter_sample(physeq, filterfun_sample(function(x) x > cutoff_value), A=filter*nsamples(physeq))
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	220
0 b4606394e7ec planemo upload simon-gladman parents: diff changeset	221 ### phyloseq object after filtered
b4606394e7ec planemo upload simon-gladman parents: diff changeset	222 physeq_filter = prune_taxa(physeq_temp, physeq)
2 20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	223
0 b4606394e7ec planemo upload simon-gladman parents: diff changeset	224 ## Transform to even sampling depth
b4606394e7ec planemo upload simon-gladman parents: diff changeset	225 physeq_filter = transform_sample_counts(physeq_filter, function(x) 1E6 * x/sum(x))
b4606394e7ec planemo upload simon-gladman parents: diff changeset	226
2 20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	227
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	228
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	229 if(plottype == 1){
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	230
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	231 # kingdom_str <- as.numeric(kingdom_str)
0 b4606394e7ec planemo upload simon-gladman parents: diff changeset	232 ## Keep only the most abundant five phyla
2 20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	233
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	234 phylum.sum = tapply(taxa_sums(physeq_filter), tax_table(physeq_filter)[, kingdom_str], sum,na.rm=TRUE)
0 b4606394e7ec planemo upload simon-gladman parents: diff changeset	235 ### number of most abundance phyla to keep
b4606394e7ec planemo upload simon-gladman parents: diff changeset	236 topphyla = names(sort(phylum.sum, TRUE))[1:keep]
2 20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	237
0 b4606394e7ec planemo upload simon-gladman parents: diff changeset	238 physeq_filter = prune_taxa((tax_table(physeq_filter)[, kingdom_str] %in% topphyla), physeq_filter)
b4606394e7ec planemo upload simon-gladman parents: diff changeset	239
2 20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	240 }else{
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	241
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	242 otu_table(physeq_filter)<-otu_table(physeq_filter)[,colSums(otu_table(physeq_filter)) > 1]
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	243
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	244 }
0 b4606394e7ec planemo upload simon-gladman parents: diff changeset	245 ### select category to plot NMDS
b4606394e7ec planemo upload simon-gladman parents: diff changeset	246 category_input = get_variable(physeq_filter, category_type) %in% category_option
b4606394e7ec planemo upload simon-gladman parents: diff changeset	247 sample_data(physeq_filter)$category_input <- factor(category_input)
b4606394e7ec planemo upload simon-gladman parents: diff changeset	248
b4606394e7ec planemo upload simon-gladman parents: diff changeset	249 physeq_ord<-ordinate(physeq_filter,method,distance)
b4606394e7ec planemo upload simon-gladman parents: diff changeset	250
2 20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	251 if(plottype == 1){
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	252 #kingdom_str = colnames(tax_table)[2]
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	253 create_OTU_PDF(pdffile,physeq_filter,physeq_ord,kingdom_str,htmlfile,pngfile_nmds,pngfile_nmds_facet)
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	254 }else{
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	255 create_SAMPLE_PDF(pdffile,physeq_filter,physeq_ord,htmlfile,pngfile_nmds,category_type)
20adf95eb758 Updated phyloseq_nmds.xml to allow for new datatype. simon-gladman parents: 1 diff changeset	256 }
0 b4606394e7ec planemo upload simon-gladman parents: diff changeset	257

Mercurial > repos > simon-gladman > phyloseq_nmds

annotate phyloseq_nmds.R @ 2:20adf95eb758 draft