This is a test version of Biostars. For the public version, visit https://www.biostars.org.
Discover novel transcripts

Hello everyone,

I am working on a project aimed at reconstructing the RNAseq transcriptome (59 bp, paired-end) to discover new transcripts in mice. I have six different tissues, and for each tissue, I have two WT replicates and two Mutant replicates. I will briefly describe my transcript assembly method.

1- Transcript reconstruction with StringTie

stringtie -p 8 -c 2 -j 2 "$BAM_FILE" -G $GTF_FILE -o "$OUTPUT_GTF"

2- Merging assembly files with StringTie, with and without using a reference

stringtie --merge -G $GTF_FILE -o $OUT_DIR/assembly_merged.gtf $INPUT_DIR/assembly_Stringtie*.gtf
stringtie --merge -o $OUT_DIR/assembly_merged.gtf $INPUT_DIR/assembly_Stringtie*.gtf

3- Transcript quantification

stringtie -p 12 -e -B $BAM_FILE -G $GTF_merged -o $OUTPUT_GTF

Questions:

  • What do you think of this approach? ( Is the method used suitable for discovering new transcripts? Are there any steps or parameters that could be optimized to improve the results?)
  • Should I merge the replicates before assembly (Step 1)?
  • How should I choose between merging with or without a reference (Step 2)?
transcriptomic assembly stringtie stringtie--merge
• 1,511 views • link
updated by Istvan Albert Istvan Albert • written by Emy Alade • Emy Alade

59 bp reads ?? auwch :-) , that's ancient data, is that possible?

and on a more constructive note: it's quite hard to get something meaningful with reads of that size I'm afraid (especially in assembly context)

• 0 views • link
written by lieven.sterck lieven.sterck

Yes, I have RNA-seq data with 59 bp read length.

So, do you think it’s not possible with this read length? enter image description here

• 0 views • link
written by Emy Alade • Emy Alade

it is possible (technically speaking) but I fear a bit the results might be disappointing though

• 0 views • link
written by lieven.sterck lieven.sterck

1 answer

Take your transcripts and align them to the genome, and see what you get.

I would venture to say that you should merge the replicates. That way transcripts that are rare in each sample would be able to be detected because you've increased their coverage by adding samples together.

I don't know what the effect of reference is on merging, it should help, frankly, as long as the reference is of decent quality.

• 0 views • link
written by Istvan Albert Istvan Albert

Log in to answer this question.

More posts like this