Viral genome vs transcripts

[vijesh]

i have a fasta file containing all the virus genome which will affect cassava. i have transcript file also in fasta. in transcript file each transcript have 100 nucleotide length. I want to know whether these transcripts are inside the virus genom. any one help me pls?

[dpryan]

One relatively straight-forward way to attack this problem would be to make a combined fasta file for the viral genomes and cassava and then map the transcripts/transcript fragments against it. You can then get an idea how likely each fragment is to come from the host or virus. I assume that tophat or something like that would be an appropriate aligner, since presumably you have a mixture of host (i.e. spliced) and viral (I assume transcripts would be single exon, but I've only made viruses, not studied them) reads and not doing so might bias things.

I should note that I've never done what you're trying to do, but if no one replies with a better idea then this is enough to get you started.

[vijesh]

Originally Posted by dpryan

One relatively straight-forward way to attack this problem would be to make a combined fasta file for the viral genomes and cassava and then map the transcripts/transcript fragments against it. You can then get an idea how likely each fragment is to come from the host or virus. I assume that tophat or something like that would be an appropriate aligner, since presumably you have a mixture of host (i.e. spliced) and viral (I assume transcripts would be single exon, but I've only made viruses, not studied them) reads and not doing so might bias things.

I should note that I've never done what you're trying to do, but if no one replies with a better idea then this is enough to get you started.

can we put the two files as input in tophat.means that genome file in fasta and also the transcript file?

[dpryan]

Originally Posted by vijesh

can we put the two files as input in tophat.means that genome file in fasta and also the transcript file?

Sort of, I can read a few different meanings into what you wrote so I'll write some pseudo-commands just to ensure that my meaning is clear. Let us assume that your Cassava genome file is called "Cassava.fa" with an annotation file "Cassava.gtf", your virus genome file is called "Viruses.fa", and the transcripts whose origin you're interested in determining are in a file called "reads.fa". The work flow would then be something like the following:

cat Cassava.fa Viruses.fa > CombinedGenome.fa
bowtie2-build CombinedGenome.fa Combined
tophat -G Cassava.gtf Combined reads.fa

You can then see which reads originated from which chromosome/contig or whichever virus by using samtools view on the resulting accepted_hits.bam file (you'll probably want to filter out reads that map equally to both Cassava and viral genomes, for which you'll probably have to write a program).

It would be good to compare the results with and without using the GTF annotation, off-hand I'm not entirely sure how or if that might bias things.

[vijesh]

i dont have the annotation file(gtf) .i have only the virus genome file (fasta) and transcript file(fasta).....

[dpryan]

You don't actually have to use the annotation file, I expect it might have just made things a bit more reliable (or maybe not, one would have to check).