how to do reference based de novo assembly using velvet

[arkilis]

As above.

Currently need to do a de novo assembly, but don't know how to use a reference in the velveth. : |


Thanks for your reply!

[arkilis]

I got some tips after google.

velveth run 31 -reference PHAC_sample_2.txt -shortPaired -fastq.gz 1_001.fastq.gz 2_001.fastq.gz &

But always got some error like:

Incomplete Sequences file (computeHSPScores)

Does anyone got any idea on this? Thanks a lot!

[mastal]

I haven't seen that error message before,

but usually when running velveth either you interleave the files containing R1 and R2, or you specify -separate if the R1 and R2 reads are in separate files.

[ctseto]

Google for "Velvet Columbus", which is the reference-guided de novo assembly pipeline.

(Though on the off-chance you get weird results)

http://bioweb2.pasteur.fr/docs/velvet/Columbus_manual.pdf

[kmcarr]

Currently need to do a de novo assembly, but don't know how to use a reference in the velveth.

Then you should probably read the Columbus_manual.pdf which is included in the velvet distribution tar file.

Basically you're doing it wrong. To do reference guided assembly in velvet you first align your reads to the reference with something like bowtie or bwa and then input the aligned read file (SAM) along with the reference file (FASTA) to velveth. You can not go directly from reads to reference guided assembly with velvet, prealignment is required.

[arkilis]

Google for "Velvet Columbus", which is the reference-guided de novo assembly pipeline.

(Though on the off-chance you get weird results)

http://bioweb2.pasteur.fr/docs/velve...bus_manual.pdf

Thanks for your advices.

[arkilis]

Then you should probably read the Columbus_manual.pdf which is included in the velvet distribution tar file.

Basically you're doing it wrong. To do reference guided assembly in velvet you first align your reads to the reference with something like bowtie or bwa and then input the aligned read file (SAM) along with the reference file (FASTA) to velveth. You can not go directly from reads to reference guided assembly with velvet, prealignment is required.

Hi I found out that if I use the -separate option, that will be fine. No such error anymore. But the N50 is pertty lame (only 100, kmer 103). I read the http://bioweb2.pasteur.fr/docs/velve...bus_manual.pdf you mentioned, but there is no clue on do alignment before velvet assembly. Why you suggest to do that? thx

[kmcarr]

I read the http://bioweb2.pasteur.fr/docs/velve...bus_manual.pdf you mentioned, but there is no clue on do alignment before velvet assembly. Why you suggest to do that? thx

On page 2, section 3 "Overview of the process", step 1 says

Map the reads against a set of target sequences (typically, an entire reference genome, made up of chromosomal sequences).

Step 4 then says

Provide this FASTA file along with the SAM/BAM alignment file

("this FASTA file means the reference file")

Pretty clear indication that alignment is the first step and you input the reference and alignment, not the read fastq files.

[arkilis]

On page 2, section 3 "Overview of the process", step 1 says

Step 4 then says ("this FASTA file means the reference file")

Pretty clear indication that alignment is the first step and you input the reference and alignment, not the read fastq files.

Genius, I did not read the manual carefully. But when I run the velveth with columbus, I got:

"WARNING: None of your read mappings recognized the reference sequence!"

Is that mean I got the wrong reference file or does not recognized at all?