using STAR to map against 100 viral species

[K]

Hello,

1) I am interested in mapping my data (unmapped fasq RNAseq files) against 100 virus species (meaning I have 100 fasta files, each having its own gff file)

Would I need to 

generate reference index files for each fasta file , and then do alignment against reference index ?

OR 

Is it possible to generate one reference STAR index file for all my 100 viruses, so that I only need to perform alignment once ?

2) As a test run I mapped my human trnascriptome file to human reference and have unmapped fastq file. 

I then took one virus (hepatitis C) fasta and gff file and then generated STAR reference genome for that.

Now I am trying to map the unmapped fastq file with this hepatitis C genome reference. Mapping seems very slow and progress.out has only the header.

I saw this post about https://groups.google.com/forum/#!msg/rna-star/cLpf7BuDnGY/nLXTE_pHDHgJ merging human + viral fasta file, but that is not something I am interested in doing. I was wondering if I can get this STAR mapping to work just on the viral reference file.

Log file attached. Any help appreciated.

Thanks,

K

[Alexander Dobin]

Hi K,

you can generate the genome for all the 100 genomes together, and perform the alignment once. Of course, you will have to deal with multi-mappers, since many reads will likely map to homologous viral sequences.

I believe the problem in your mapping was caused a bug in the genome generation without annotation. I have fixed it in the patch on github-master: https://github.com/alexdobin/STAR/archive/master.zip

Please try it out and let me know if it solved the problem.

If you are mapping not purely viral RNA, but rather a mix of human and viral RNA, I would recommend generating a combined genome of human and 100 viruses. This will make mapping faster, and will probably increase  mapping accuracy.

Cheers

Alex