STARlong on Pacbio data

[Fujun Qin]

Hi, Alex,

Recently I tried to run STARlong on Pacbio data from NCBI.   My goal is to align my artificial genome(short fusion RNA sequence) to the Pacbio RNA seq data to find reads supporting  my fusion RNA junction.   My artificial genome sequence is YP.fasta in the attached file. 

First, I used star.slurm( command job for our server) to generate artificial genome using my YP.fasta sequence.  

Then I used STARlong.slurm to align my genome to the Pacbio RNA-sequence(SRR1853186.fastq, which is download from NCBI with the sratool kit using "--table SEQUENCE' to dump SRR1853186).

It took several days to run, but never give me any result. could you help me out the issues ? 

Thank you ,

Fujun

[Alexander Dobin]

Hi Fujun,

this likely means that there were no good matches to your sequence.

I would recommend that you generate a combined genome, human reference and YP.

Then you will see how many reads can be mapped to the reference.

Also, mapping to the combined genome will be faster than to YP alone.

Cheers

Alex

[Fujun Qin]

Thank you. Alex.

two more questions on what you mentioned.

1. how should I generate the combined genome ?  directly paste the human transcriptom to my YP, then generate combined genome ? or directly paste the human DNA sequence to my YP?

2. Are those command I used ok ?

Thanks,

Fujun

[Alexander Dobin]

Hi Fujun,

1. You can input two fasta files at the genome generation step:

$ STAR --runMode genomeGenerate --genomeDir ./ --genomeFastaFiles HumanGenome.fasta YP.fasta --sjdbGTFfile Human.gtf --runThreadN 12

2. The commands look fine. 

Cheers

Alex

[Fujun Qin]

Thank you. I will try.

[Fujun Qin]

How much time will it take to generate the genome ? It took me more than 24hours, still didn't finish.

Thanks,

Fujun

[Alexander Dobin]

Hi Fujun,

should take a few hours. How much RAM do you have? You would need ~32GB for the whole human genome.

Cheers

Alex

[Fujun Qin]

Thank you. I have enough RAM.  I am using server.

Fujun