Dear All,
I have rna-seq data of 3 ginger sample of which are from different condition.
I have assembled them using trinity tool, hence i got three different transcript assembly.
Since public database doesn't have reference genome of ginger, therefore to build own custom transcript genome of ginger i merged and cluster all three trinity assemblies following this workflow
Workflow:-
# Merging all samples trinity assemblies
cat 0-trinity_out/Unigenes_S1.fasta 0-trinity_out/Unigenes_S3.fasta 0-trinity_out/Unigenes_S4.fasta > 0-trinity_out/Unigenes_all.fasta
# Clustring merged assemlies at 100% identity
cd-hit-est -i 0-trinity_out/Unigenes_all.fasta -o 1-cdhit_out/cdhit_clstr.fasta -c 1.0 -n 8 -g 1 -r 1 -B 1 -p 1 &> 1-cdhit_out/cdhit_clstr.log
cat 1-cdhit_out/cdhit_clstr.fasta | python /home/blab/myscript/rename_multifasta.py Seq > 2-tgicl_out/cdhit_clstr.fasta
read_fasta -i 2-tgicl_out/cdhit_clstr.fasta | write_fasta -x -o 2-tgicl_out/cdhit_clstr_singleline.fasta
mv 2-tgicl_out/cdhit_clstr_singleline.fasta 2-tgicl_out/cdhit_clstr.fasta
# Further clustring using TGICL
$TGICL -F cdhit_clstr.fasta -p 99 -l 50 -v 100 -c 3 -L -b blab:cbt@soa:mysql:127.0.0.1:tgicldb -R
cat asm_1/contigs asm_2/contigs asm_3/contigs asm_1/singlets asm_2/singlets asm_3/singlets > final_contigs.fasta
# EvidentialGene::tr2aacds.pl pipeline script for processing final_contigs.fasta into the most biologically useful "best" set of mRNA, classified into "primary" and "alternate" transcripts.
/home/blab/Downloads/evigene/scripts/prot/tr2aacds_qsub.sh
# The output is a neat pile of "okay" and "drop" transcripts
# The "okay" set is annotated using trinotate tool to used as custom transcript genome.
# tr2aacds_qsub file is attached here.
I do not work with plants but this is likely not different from what should be done with mouse.
We are merging fastqs BEFORE trinity. That way if you want to look at expression levels you would have a common reference. In addition you would have more power to reconstruct transcripts expressed at low levels.
On the flip side you may increase the noise and if those samples are very different, transcripts that are expressed at low levels in just one sample may get lost in the noise. Then perhaps a reciprocal blast may identify true pairs (I am not sure clustering would work well).
I do not work with plants but this is likely not different from what should be done with mouse.
We are merging fastqs BEFORE trinity. That way if you want to look at expression levels you would have a common reference. In addition you would have more power to reconstruct transcripts expressed at low levels.
On the flip side you may increase the noise and if those samples are very different, transcripts that are expressed at low levels in just one sample may get lost in the noise. Then perhaps a reciprocal blast may identify true pairs (I am not sure clustering would work well).
I agree with Stefan, you should merge or use all available transcripts to build your de novo transcriptome with Trinity.
This will allow you to have a more complete transcriptome that includes mRNA that may only be found in one or two of the three different samples. When comparing the expression levels you may find that each sample has some tissue-specific expression, this is especially important when building a transcriptome from samples under different environmental conditions.
Merging will be the way to go. In my experience, I keep each individual treatment and then generate a mixed one. I agree with Zacharie.
Agreed to merge different trinity transcript for assembly of a same plant. This would be your reference transcript and useful later, since you don't have reference sequence. It is good that you had removed identical transcripts by clustering process.
I see this is a rather old thread, but I came across it, so hopefully some of you are still around.
Wanted to ask - how do you merge multiple transcript sets (in fasta format) produced by Trinity? Is there a tool that can do it in a smart way that will elongate transcripts based on overlaps, collapse partial transcripts found within others etc.? I see that cd-hit-est was suggested here, but is it really the right tool for the job?
Thanks!
Hi Lior
you could give a try to https://github.com/cboursnell/transfuse. I used it in the past and it gave me good results. More recently I have found this tool (https://github.com/vsbuffalo/blast2cap3) which looks promising but I have not tried myself.
I hope this information might be useful.
Riccardo
Bioinformatics data analyst at www.sequentiabiotech.com
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Bioinformatics