convert tRNA coordinates Result to Fasta format

Question

I believe many of the experts here have gone through this issue,

I have run tRNAscan-SE on my genome to predict all the tRNAs, and have successfully gotten all the tRNA coordinates like start, end and type all arranged in a table.

My question is how can I convert this result to fasta format as long I have all the details about the tRNA ready?

Is there any tool that I can use to get the fasta sequences of the regions described in my table from the original genome?

---

Sequence                tRNA    Bounds  tRNA    Anti    Intron Bounds   Inf       
Name            tRNA #  Begin   End     Type    Codon   Begin   End     Score   Note
--------        ------  -----   ------  ----    -----   -----   ----    ------  ------
NW_011590950.1  1       36704   36632   Ala     AGC     0       0       40.8    pseudo

Answer 1

You can select “Output BED format” on the tRNAscan-SE server:

Given the output BED format and your input reference in FASTA format, you can easily generate a FASTA file of tRNA sequences via bedtools getfasta:

bedtools getfasta -name -s -fo ‹output.fasta› -fi ‹reference.fasta› -bed ‹tRNAscan-SE-output.bed›

Answer 2

You can do this using the tools from the exonerate package. I assume you have all the sequences shown in your table in a single, multifasta file. If so, first create a single file per sequence using fastaexplode:

fastaexplode reference.fa

This will create one file for each sequence in reference.fa. Now, you can read through the results table, and extract the sequences:

tmpFile=$(mktemp)
awk -F"\t" '{ 
                if(NR>3){ printf "%s %s ", $1, $2; 
                    if($4>$3){
                        print $3, $4-$3
                    }
                    else{
                        print $4,$3-$4,"-"
                    }
                }
            }' file | while read name num start len rev; do 
                fastasubseq "$name".fa $start $len > $num.fa
                if [[ ! -z $rev ]]; then 
                    fastarevcomp $num.fa > $tmpFile && 
                    mv $tmpFile $num.fa
                fi
                done

I would suggest you go with Konrad's approach, however, that looks much cleaner and simpler.