Merge pull request #65 from awasyn/depacc2lin

Rename functions in acc2lin - fixes #35
JRaviLab · Oct 8, 2024 · 3d8ec75 · 3d8ec75
2 parents 4a4efa6 + 2d47952
commit 3d8ec75
Show file tree

Hide file tree

Showing 8 changed files with 71 additions and 70 deletions.
diff --git a/NAMESPACE b/NAMESPACE
@@ -4,15 +4,16 @@ export(BinaryDomainNetwork)
 export(DownloadAssemblySummary)
 export(GCA2lin)
 export(GenContextNetwork)
+export(IPG2Lineage)
 export(LineagePlot)
 export(RepresentativeAccNums)
 export(acc2FA)
+export(acc2Lineage)
 export(acc2fa)
-export(acc2lin)
 export(addLeaves2Alignment)
+export(addLineage)
 export(addName)
 export(add_leaves)
-export(add_lins)
 export(add_name)
 export(add_tax)
 export(advanced_opts2est_walltime)
@@ -40,7 +41,7 @@ export(create_all_col_params)
 export(create_lineage_lookup)
 export(create_one_col_params)
 export(domain_network)
-export(efetch_ipg)
+export(efetchIPG)
 export(extractAccNum)
 export(filter_by_doms)
 export(filter_freq)
@@ -57,7 +58,6 @@ export(get_accnums_from_fasta_file)
 export(get_job_message)
 export(get_proc_medians)
 export(get_proc_weights)
-export(ipg2lin)
 export(ipr2viz)
 export(ipr2viz_web)
 export(lineage.DA.plot)
@@ -73,6 +73,7 @@ export(map_advanced_opts2procs)
 export(msa_pdf)
 export(plot_estimated_walltimes)
 export(prot2tax)
+export(prot2tax_old)
 export(removeAsterisks)
 export(removeEmptyRows)
 export(removeTails)
@@ -86,7 +87,7 @@ export(run_rpsblast)
 export(selectLongestDuplicate)
 export(send_job_status_email)
 export(shorten_lineage)
-export(sink.reset)
+export(sinkReset)
 export(stacked_lin_plot)
 export(summ.DA)
 export(summ.DA.byLin)

diff --git a/R/acc2lin.R b/R/acc2lin.R
@@ -14,16 +14,16 @@
 #'
 #' @examples
 #' \dontrun{
-#' sink.reset()
+#' sinkReset()
 #' }
-sink.reset <- function() {
+sinkReset <- function() {
     for (i in seq_len(sink.number())) {
         sink(NULL)
     }
 }
 
 
-#' Add Lineages
+#' addLineage
 #'
 #' @param df
 #' @param acc_col
@@ -41,13 +41,13 @@ sink.reset <- function() {
 #'
 #' @examples
 #' \dontrun{
-#' add_lins()
+#' addLineage()
 #' }
-add_lins <- function(df, acc_col = "AccNum", assembly_path,
-    lineagelookup_path, ipgout_path = NULL, plan = "sequential") {
+addLineage <- function(df, acc_col = "AccNum", assembly_path,
+    lineagelookup_path, ipgout_path = NULL, plan = "sequential", ...) {
     s_acc_col <- sym(acc_col)
     accessions <- df %>% pull(acc_col)
-    lins <- acc2lin(accessions, assembly_path, lineagelookup_path, ipgout_path, plan)
+    lins <- acc2Lineage(accessions, assembly_path, lineagelookup_path, ipgout_path, plan)
 
     # Drop a lot of the unimportant columns for now? will make merging much easier
     lins <- lins[, c(
@@ -64,11 +64,11 @@ add_lins <- function(df, acc_col = "AccNum", assembly_path,
 }
 
 
-#' acc2lin
+#' acc2Lineage
 #'
 #' @author Samuel Chen, Janani Ravi
 #'
-#' @description This function combines 'efetch_ipg()' and 'ipg2lin()' to map a set
+#' @description This function combines 'efetchIPG()' and 'IPG2Lineage()' to map a set
 #' of protein accessions to their assembly (GCA_ID), tax ID, and lineage.
 #'
 #' @param accessions Character vector of protein accessions
@@ -85,25 +85,25 @@ add_lins <- function(df, acc_col = "AccNum", assembly_path,
 #'
 #' @examples
 #' \dontrun{
-#' acc2lin()
+#' acc2Lineage()
 #' }
-acc2lin <- function(accessions, assembly_path, lineagelookup_path, ipgout_path = NULL, plan = "sequential") {
+acc2Lineage <- function(accessions, assembly_path, lineagelookup_path, ipgout_path = NULL, plan = "sequential", ...) {
     tmp_ipg <- F
     if (is.null(ipgout_path)) {
         tmp_ipg <- T
         ipgout_path <- tempfile("ipg", fileext = ".txt")
     }
-    efetch_ipg(accessions, out_path = ipgout_path, plan)
+    efetchIPG(accessions, out_path = ipgout_path, plan)
 
-    lins <- ipg2lin(accessions, ipgout_path, assembly_path, lineagelookup_path)
+    lins <- IPG2Lineage(accessions, ipgout_path, assembly_path, lineagelookup_path)
 
     if (tmp_ipg) {
         unlink(tempdir(), recursive = T)
     }
     return(lins)
 }
 
-#' efetch_ipg
+#' efetchIPG
 #'
 #' @author Samuel Chen, Janani Ravi
 #'
@@ -123,12 +123,12 @@ acc2lin <- function(accessions, assembly_path, lineagelookup_path, ipgout_path =
 #'
 #' @examples
 #' \dontrun{
-#' efetch_ipg()
+#' efetchIPG()
 #' }
-efetch_ipg <- function(accnums, out_path, plan = "sequential") {
+efetchIPG <- function(accnums, out_path, plan = "sequential", ...) {
     if (length(accnums) > 0) {
         partition <- function(in_data, groups) {
-            # \\TODO This function should be defined outside of efetch_ipg(). It can be non-exported/internal
+            # \\TODO This function should be defined outside of efetchIPG(). It can be non-exported/internal
             # Partition data to limit number of queries per second for rentrez fetch:
             # limit of 10/second w/ key
             l <- length(in_data)
@@ -168,7 +168,7 @@ efetch_ipg <- function(accnums, out_path, plan = "sequential") {
     }
 }
 
-#' ipg2lin
+#' IPG2Lineage
 #'
 #' @author Samuel Chen, Janani Ravi
 #'
@@ -191,10 +191,10 @@ efetch_ipg <- function(accnums, out_path, plan = "sequential") {
 #'
 #' @examples
 #' \dontrun{
-#' ipg2lin()
+#' IPG2Lineage()
 #' }
 #'
-ipg2lin <- function(accessions, ipg_file, assembly_path, lineagelookup_path) {
+IPG2Lineage <- function(accessions, ipg_file, assembly_path, lineagelookup_path, ...) {
     ipg_dt <- fread(ipg_file, sep = "\t", fill = T)
 
     ipg_dt <- ipg_dt[Protein %in% accessions]
@@ -211,7 +211,7 @@ ipg2lin <- function(accessions, ipg_file, assembly_path, lineagelookup_path) {
 
 
 
-# efetch_ipg <- function(accnums, outpath)
+# efetchIPG <- function(accnums, outpath)
 # {
 #   SIZE = 250
 #   lower_bound = 1

diff --git a/R/lineage.R b/R/lineage.R
@@ -133,7 +133,7 @@ GCA2lin <- function(prot_data,
 ###################################
 ## !! @SAM why is this called lins?
 ###################################
-#' add_lins
+#' addLineage
 #'
 #' @param df
 #' @param acc_col
@@ -149,11 +149,11 @@ GCA2lin <- function(prot_data,
 #' @export
 #'
 #' @examples
-add_lins <- function(df, acc_col = "AccNum", assembly_path,
+addLineage <- function(df, acc_col = "AccNum", assembly_path,
     lineagelookup_path, ipgout_path = NULL, plan = "multicore") {
     acc_sym <- sym(acc_col)
     accessions <- df %>% pull(acc_sym)
-    lins <- acc2lin(accessions, assembly_path,
+    lins <- acc2Lineage(accessions, assembly_path,
         lineagelookup_path, ipgout_path,
         plan = plan
     )
@@ -178,13 +178,13 @@ add_lins <- function(df, acc_col = "AccNum", assembly_path,
 #######################################
 ## Map Protein Accessions to Lineage ##
 #######################################
-#' acc2lin
+#' acc2Lineage
 #'
 #' @description
 #' Function to map protein accession numbers to lineage
 #'
 #' @author Samuel Chen, Janani Ravi
-#' @description This function combines 'efetch_ipg()' and 'ipg2lin()' to map a set
+#' @description This function combines 'efetchIPG()' and 'IPG2Lineage()' to map a set
 #' of protein accessions to their assembly (GCA_ID), tax ID, and lineage.
 #'
 #' @param accessions Character vector of protein accessions
@@ -200,17 +200,17 @@ add_lins <- function(df, acc_col = "AccNum", assembly_path,
 #' @export
 #'
 #' @examples
-acc2lin <- function(accessions, assembly_path, lineagelookup_path,
+acc2Lineage <- function(accessions, assembly_path, lineagelookup_path,
     ipgout_path = NULL, plan = "multicore") {
     tmp_ipg <- F
 
     if (is.null(ipgout_path)) {
         tmp_ipg <- T
         ipgout_path <- tempfile("ipg", fileext = ".txt")
     }
-    efetch_ipg(accessions, out_path = ipgout_path, plan = plan)
+    efetchIPG(accessions, out_path = ipgout_path, plan = plan)
 
-    lins <- ipg2lin(accessions, ipgout_path, assembly_path, lineagelookup_path)
+    lins <- IPG2Lineage(accessions, ipgout_path, assembly_path, lineagelookup_path)
 
     # if(tmp_ipg)
     # {
@@ -227,7 +227,7 @@ acc2lin <- function(accessions, assembly_path, lineagelookup_path,
 #########################################
 ## Download IPG results for Accessions ##
 #########################################
-#' efetch_ipg
+#' efetchIPG
 #'
 #' @author Samuel Chen, Janani Ravi
 #' @description Perform efetch on the ipg database and write the results to out_path
@@ -245,7 +245,7 @@ acc2lin <- function(accessions, assembly_path, lineagelookup_path,
 #' @export
 #'
 #' @examples
-efetch_ipg <- function(accessions, out_path, plan = "multicore") {
+efetchIPG <- function(accessions, out_path, plan = "multicore") {
     if (length(accessions) > 0) {
         partition <- function(v, groups) {
             # Partition data to limit number of queries per second for rentrez fetch:
@@ -295,7 +295,7 @@ efetch_ipg <- function(accessions, out_path, plan = "multicore") {
 #########################################
 ## Maps IPG results to TaxID + Lineage ##
 #########################################
-#' ipg2lin
+#' IPG2Lineage
 #'
 #' @author Samuel Chen, Janani Ravi
 #' @description Takes the resulting file of an efetch run on the ipg database and
@@ -317,7 +317,7 @@ efetch_ipg <- function(accessions, out_path, plan = "multicore") {
 #' @export
 #'
 #' @examples
-ipg2lin <- function(accessions, ipg_file,
+IPG2Lineage <- function(accessions, ipg_file,
     refseq_assembly_path, genbank_assembly_path,
     lineagelookup_path) {
     ipg_dt <- fread(ipg_file, sep = "\t", fill = T)

diff --git a/man/ipg2lin.Rd → man/IPG2Lineage.Rd b/man/ipg2lin.Rd → man/IPG2Lineage.Rd
diff --git a/man/acc2lin.Rd → man/acc2Lineage.Rd b/man/acc2lin.Rd → man/acc2Lineage.Rd
diff --git a/man/add_lins.Rd → man/addlineage.Rd b/man/add_lins.Rd → man/addlineage.Rd