GO results

[wjq1981]

Hi, when I use the command "deepgoplus --data-root /hdd/Software/deepgoplus/data/ --in-file test.fasta ---out-file out.file" the result is just an id number, how can I get the corresponding How can I get the corresponding GO number?

[Rohit-Satyam]

Hi @wjq1981

If I understood correctly you want to convert GO IDs to Description right? If so here is a little trick in R to get corresponding GO terms that I tried yesterday:

library(tidyverse)
## Will read tsv as 19 column Dataframe
deepgoplus <- read_tsv("deepgoplus_results.tsv", col_names = FALSE)
## Replacing the last colum tab separated values to semicolon separated values
deepgoplus[,ncol(deepgoplus)] <- gsub("\t",";", deepgoplus$X19)

## Aggregate columns into a single colum
deepgoplus <- unite(deepgoplus, col='go_term', c(2:ncol(deepgoplus)), sep=';')

## Converting the GO terms per row to corresponding description using pRoloc::goidToTerm function
invisible(lapply(1:nrow(deepgoplus), function(x){

## Taking one row at a time and converting it to dataframe
    tempdf<-data.frame(go=(stringr::str_split(deepgoplus[x,2], pattern = "[;]")[[1]])) %>%
    separate(go, into = c('go', 'conf'), sep = "\\|")
	
## Converting GO IDs to description
    tempdf$desc <- pRoloc::goIdToTerm(tempdf$go, names = TRUE, keepNA = TRUE)

## At this point you can also subset most meaningful and high confidence GO terms
    tempdf<-subset(tempdf, conf == max(conf))

## Update the second column with the high confidence go term description
  deepgoplus[x,2] <<- paste(tempdf$desc, tempdf$conf, sep="|",collapse = ";")
}))

Now I would like to mention that most of the high confidence GO terms provided by deepgoplus are useless like in screenshot show below and are not very informative when working with hypothetical protein.

I also tried to update the GO database, but the script keeps on failing as well the following error

./update.py
Checking new release date...
ERROR:root:'x-uniprot-release'

[wjq1981]

Hi @wjq1981

If I understood correctly you want to convert GO IDs to Description right? If so here is a little trick in R to get corresponding GO terms that I tried yesterday:

library(tidyverse)
## Will read tsv as 19 column Dataframe
deepgoplus <- read_tsv("deepgoplus_results.tsv", col_names = FALSE)
## Replacing the last colum tab separated values to semicolon separated values
deepgoplus[,ncol(deepgoplus)] <- gsub("\t",";", deepgoplus$X19)

## Aggregate columns into a single colum
deepgoplus <- unite(deepgoplus, col='go_term', c(2:ncol(deepgoplus)), sep=';')

## Converting the GO terms per row to corresponding description using pRoloc::goidToTerm function
invisible(lapply(1:nrow(deepgoplus), function(x){

## Taking one row at a time and converting it to dataframe
    tempdf<-data.frame(go=(stringr::str_split(deepgoplus[x,2], pattern = "[;]")[[1]])) %>%
    separate(go, into = c('go', 'conf'), sep = "\\|")
	
## Converting GO IDs to description
    tempdf$desc <- pRoloc::goIdToTerm(tempdf$go, names = TRUE, keepNA = TRUE)

## At this point you can also subset most meaningful and high confidence GO terms
    tempdf<-subset(tempdf, conf == max(conf))

## Update the second column with the high confidence go term description
  deepgoplus[x,2] <<- paste(tempdf$desc, tempdf$conf, sep="|",collapse = ";")
}))

Now I would like to mention that most of the high confidence GO terms provided by deepgoplus are useless like in screenshot show below and are not very informative when working with hypothetical protein.

I also tried to update the GO database, but the script keeps on failing as well the following error

./update.py
Checking new release date...
ERROR:root:'x-uniprot-release'

Thank you so much for your help. I will give it a try.