QC of 454 reads

[flobpf]

Hi,
I wonder how people analyze the quality of the 454 Titanium reads? What softwares/tools/pipelines exist? I was thinking of converting the SFF files to FASTQ and then doing the regular fastx boxplots. But I wonder whether there are any tools that can directly assess the quality from the SFF files and whether my approach above is a good one to take.

Thanks for any suggestions!

[JackieBadger]

Hi,
I wonder how people analyze the quality of the 454 Titanium reads? What softwares/tools/pipelines exist? I was thinking of converting the SFF files to FASTQ and then doing the regular fastx boxplots. But I wonder whether there are any tools that can directly assess the quality from the SFF files and whether my approach above is a good one to take.

Thanks for any suggestions!

Hey,

You may get a quicker and better answer if you post under the "bioinformatics" thread.
See http://seqanswers.com/forums/showthread.php?t=10784 for QC

Additional, watch all the video demos at the Galaxy pipeline site (at bottom of the above link)

I have run some trials of converting FASTA to FASTX for quality assessments. No good. Basically the Illumina QC thresholds and scoring are quite different from 454, thus the distributions may be pretty hard to interpret and may be incorrect. Use the PRINSEQ program in the above thread.

In regards to your question on pipelines....what data do you have? what are your questions?

J

[flobpf]

PRINSEQ works

Hey,

You may get a quicker and better answer if you post under the "bioinformatics" thread.
See http://seqanswers.com/forums/showthread.php?t=10784 for QC

Additional, watch all the video demos at the Galaxy pipeline site (at bottom of the above link)

I have run some trials of converting FASTA to FASTX for quality assessments. No good. Basically the Illumina QC thresholds and scoring are quite different from 454, thus the distributions may be pretty hard to interpret and may be incorrect. Use the PRINSEQ program in the above thread.

In regards to your question on pipelines....what data do you have? what are your questions?

J

Hi JackieBadger,

Thanks for your reply. I just tried PRINSEQ and the plots the web version generates for 454 reads are a lot better than FASTQC. Thanks for pointing me to it. Standalone works fine too.

About the pipeline, I meant a pipeline to assess quality of 454 reads. But later on, I intend to use Newbler/Abyss for assembly etc.

[JackieBadger]

Hi JackieBadger,

Thanks for your reply. I just tried PRINSEQ and the plots the web version generates for 454 reads are a lot better than FASTQC. Thanks for pointing me to it. Standalone works fine too.

About the pipeline, I meant a pipeline to assess quality of 454 reads. But later on, I intend to use Newbler/Abyss for assembly etc.

Cool.

I have heard Velvet/Abyss are good, but not too sure on the algorithm.
Look into MIRA3...especially good for EST assemblies. Their algorithm is particularly appealing a it uses quality scores in the alignment process.

I have started to use MIRA and am very impressed with it's SNP calling abilities and handling of the data.

Good luck.

[tonybolger]

I have heard Velvet/Abyss are good, but not too sure on the algorithm.

They are de-bruijn graph based, and will likely not work well on 454 data.

Look into MIRA3...especially good for EST assemblies. Their algorithm is particularly appealing a it uses quality scores in the alignment process.

MIRA (if the data size is relatively small - it's slow as hell) or especially Newbler would be likely the best choices.