I have elaborated a de novo transcriptome assembly by means of Illumina sequencing output with more than 300 million reads. I am now validating some gene models and looking for new isoforms in that assembly using blastn searches. I take into account the sequence identity and the coverage of the blast output to validate gene models. In some cases, I find a match between my sequence and one transcript from the assembly (sequence identity >95% and sequence coverage >90%), which confirm that my gene model would be correct and the expression of this gene in the studied conditions. However, in some cases I find a perfect match but it is partial e.g. the transcript from the assembly covers only 50% of my sequence. Could I consider this gene as expressed in the experimental conditions?
Thanks
Jose
Jose,
You don't state what you are validating the de novo transcriptome assembly against. If this is not complete (fragmented genome assembly) or you are using gene predictions which may be incorrect then you could get these kind of partial hits. Further is it possible that you have chimeric transcripts (artifacts of the assembly) which again could account for a partial match. If you want to discuss this more then mail me
Dan
Jose, conozco un par de expertos en la UV que te pueden ayudar. Si todavia te interesa, enviame un mail.
Jose,
I would not advise you to use blast for validating the transcriptome assembly. Its better to use Bowtie or BWA. This will create SAM file, which can be parsed to calculate the coverage of each transcript and its isoform.
Thanks all for your comments.
Dear Suyash, I am also using Tophat pipeline to validate gene models.
Regards
Jose
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Bioinformatics