I check whether the transcript overlaps more than one gene; if that happens and the following conditions apply to two or more genes, it is a fusion:

 - junction in common

- at least 10% of the reference gene is covered (cDNA, not genomic)

In case you dont have a reference : You should align the RNASeq reads to the putative transcripts and see how the depth of coverage is across the length of transcript. In any assembled transcript which has huge coverage variation at one end as compare to the other end. Or may be both the ends have good deep coverage but are separated by poor coverage. Both the cases can be classed as a chimeric region.

Hey all, I know there are some tools to delete Chimera in Qiime like - Usearch or Chimera slayer but I can't understand how can I run it before the qiime commend in the terminal. when I try to run the commend I get a technical error.

thanks in advance, Liran 

Chimera deletion

export PATH=$PATH:/path/to/usearch61/bin/

export PATH=$PATH:/path/to/qiime 

Step 1: identify chimeras

identify_chimeric_seqs.py -i seqs.fna -m usearch61 -o usearch_checked_chimeras/ -r Chimera_db/uchime_chimera.fa 

Step 2: delete chimeras

filter_fasta.py -f seqs.fna -o seqs_chimeras_filtered.fna -s usearch_checked_chimeras2/chimeras.txt -n 

Cheers

Shab

Hello Bukola,

To partly answer your question, I’d like to think of chimera as belonging to two categories: synthetic (assembly-derived) and natural (such as those in certain cancer or normal human cell).  With a ‘good’ genome assembly on which to align/map the transcripts/RNASeq reads and summarising the alignment could indicate which of your sequences are chimeric e.g Dissect.  For non-model organisms, you could align the transcripts/RNASeq reads to a proteome or tanscriptome e.g https://bitbucket.org/yangya/optimize_assembler. You may also want to look at this review http://onlinelibrary.wiley.com/wol1/doi/10.1002/wrna.1382/full

 In the second part of your question, it appears to me you have generated a consensus sequence from a clustering tool. For example if you have used CDhit for this purpose, you should be able to locate the intermediate file which has the cluster relationship of each sequence.

 HTH