Take "complete" transcripts from Transdecoder and parse the GTF file using a FASTA file that has a subset of the transcripts in the GTF file. In the process, GTF_parser.pl will also remove entries that have formatting issues, so that the output GTF should be a clean output for SnpEff's build database command. I am not sure why Transdecoder has these errors, but sometimes transcripts will have:
- Transcripts ending in the GTF file after the end of the chromosome/scaffold/contig
- Start codons that are not three bases long
- Stop codons that are not three bases long
Also, SnpEff doesn't like having the "Gene" line in the GTF file for multiple transcripts. SnpEff also doesn't seem to like to have the annotations for the 5'UTR, 3'UTR, start codon and stop codon. I was running SnpEffv4.3t for this work, so subsequent versions may handle these annotations differently. However, GTF_parser.pl cleans all of these issues for at least SnpEffv4.3t.
An example of a command to use GTF_parser.pl is:
perl GTF_parser.pl --gtf_file=gtf_file.gtf --transcriptome_input=transcriptome_input.fasta --genome_input=genome_input.fasta --output_file=output_file.gtf
The --gtf_input option accepts the GTF output of Transdecoder after it has been filtered by transcripts annotated as "complete".
The --transcriptome_input option accepts your subsetted transcriptome FASTA file. I think (while I haven't explicitly tested this) if you have a transcriptome FASTA that contains all of the transcripts entries in the GTF input, it will still clean up the file and remove the problematic GTF entries and superfluous annotations that SnpEff doesn't like.
The --genome_input option accepts the genome FASTA file. We just need this so we know if transcript annotations run off of the chromosome/scaffold/contig.
The --output_file option accepts the cleaned GTF output filename.
An example of a GTF file input from Transdecoder might have entries that look like:
C1174700 transdecoder 5UTR 1077 1197 0 - 0 gene_id "CUFF.9826"; transcript_id "CUFF.9826.2.p2"; Name "ORF type:complete len:123 (+),score=29.48";
C1174700 transdecoder start_codon 1074 1076 0 - 0 gene_id "CUFF.9826"; transcript_id "CUFF.9826.2.p2"; Name "ORF type:complete len:123 (+),score=29.48";
C1174700 transdecoder CDS 855 1076 0 - 0 gene_id "CUFF.9826"; transcript_id "CUFF.9826.2.p2"; Name "ORF type:complete len:123 (+),score=29.48";
C1174700 transdecoder exon 855 1197 0 - . gene_id "CUFF.9826"; transcript_id "CUFF.9826.2.p2"; Name "ORF type:complete len:123 (+),score=29.48";
C1174700 transdecoder CDS 636 779 0 - 0 gene_id "CUFF.9826"; transcript_id "CUFF.9826.2.p2"; Name "ORF type:complete len:123 (+),score=29.48";
C1174700 transdecoder stop_codon 633 635 0 - 0 gene_id "CUFF.9826"; transcript_id "CUFF.9826.2.p2"; Name "ORF type:complete len:123 (+),score=29.48";
C1174700 transdecoder 3UTR 628 632 0 - 0 gene_id "CUFF.9826"; transcript_id "CUFF.9826.2.p2"; Name "ORF type:complete len:123 (+),score=29.48";
C1174700 transdecoder exon 628 779 0 - . gene_id "CUFF.9826"; transcript_id "CUFF.9826.2.p2"; Name "ORF type:complete len:123 (+),score=29.48";
C1174700 transdecoder 3UTR 2 563 0 - 0 gene_id "CUFF.9826"; transcript_id "CUFF.9826.2.p2"; Name "ORF type:complete len:123 (+),score=29.48";
C1174700 transdecoder exon 2 563 0 - . gene_id "CUFF.9826"; transcript_id "CUFF.9826.2.p2"; Name "ORF type:complete len:123 (+),score=29.48";
C1174700 transdecoder transcript 2 1197 0 - . gene_id "CUFF.9826"; transcript_id "CUFF.9826.2.p2"; Name "ORF type:complete len:123 (+),score=29.48";
C1174700 transdecoder gene 2 1197 0 - . gene_id "CUFF.9826"; Name "ORF type:complete len:123 (+),score=29.48";
Notice how this entry has type:complete. In order to parse out the "complete" transcripts prior to running GTF_parser.pl, I just ran:
grep "type:complete" transcoder_output.gtf > gtf_file.gtf
After GTF_parser.pl, this entry will look like:
C1174700 transdecoder gene 2 1197 0 - . gene_id "CUFF.9826"; Name "ORF type:complete len:123 (+),score=29.48";
C1174700 transdecoder transcript 2 1197 0 - . gene_id "CUFF.9826"; transcript_id "CUFF.9826.2.p2"; Name "ORF type:complete len:123 (+),score=29.48";
C1174700 transdecoder CDS 855 1076 0 - 0 gene_id "CUFF.9826"; transcript_id "CUFF.9826.2.p2"; Name "ORF type:complete len:123 (+),score=29.48";
C1174700 transdecoder exon 855 1197 0 - . gene_id "CUFF.9826"; transcript_id "CUFF.9826.2.p2"; Name "ORF type:complete len:123 (+),score=29.48";
C1174700 transdecoder CDS 636 779 0 - 0 gene_id "CUFF.9826"; transcript_id "CUFF.9826.2.p2"; Name "ORF type:complete len:123 (+),score=29.48";
C1174700 transdecoder exon 628 779 0 - . gene_id "CUFF.9826"; transcript_id "CUFF.9826.2.p2"; Name "ORF type:complete len:123 (+),score=29.48";
C1174700 transdecoder exon 2 563 0 - . gene_id "CUFF.9826"; transcript_id "CUFF.9826.2.p2"; Name "ORF type:complete len:123 (+),score=29.48";
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent