using sff_extract

[veronique]

being a newbie, I have a very simple question:

I am trying to convert sff formatted 454 data into fasta, fasta quality files and an xml files using sff_extract in python2.6 using the following command:

sff_extract -s proj_in.454.fasta -q proj_in.454.fasta.qual -x proj_traceinfo_in.454.xml FX3UAMY01.sff

this results in a syntax error. Does anyone have a suggestion what's wrong?

Thanks, Veronique

[AnthonyB]

Hi,

AFAIK you only use the -s, -q or -x options if you want only one of the files. To get all three you only need to use the -o option to set the output name (if you want) and then the names of the sff files. You have to manually rename the xml file afterwards however. e.g "sff_extract -o proj_in.454" should output "proj_in.454.fasta, proj_in.454.fasta.qual and proj_in.454.xml"

[Torst]

I just use the "sffinfo" tool that 454 distribute.

sffinfo -s file.sff > file.fasta
sffinfo -q file.sff > file.qual
sffinfo -m file.sff > file.manifest.xml

Here is the usage:

Usage: sffinfo [options...] [- | sfffile] [accno...]
Options:
-s or -seq Output just the sequences
-q or -qual Output just the quality scores
-f or -flow Output just the flowgrams
-t or -tab Output the seq/qual/flow as tab-delimited lines
-n or -notrim Output the untrimmed sequence or quality scores
-m or -mft Output the manifest text

[Jose Blanca]

Which is the error message?

[foolishbrat]

Is there a way I can download "sffinfo". I tried to search google but of no avail.

[Jose Blanca]

I think sffinfo is not free software.

[AnthonyB]

I think you can only obtain sffinfo as part of the sfftools package that is supplied by Roche to groups that have the 454 sequencing machines. I'm pretty sure that this was the basis behind the production of sff_extract as many people cannot easily get access to the 454 software to extract the 454 reads from sff files.

[maubp]

I think sffinfo is not free software.

However, Roche are generally relaxed about giving end users access, see:

http://seqanswers.com/forums/showthread.php?t=114

Note that the Newbler tools (sffinfo, plus the assembler and read mapper) are for Linux only.

[chayan]

I am new to this. I want to convert a iontorrent sff to fastq and xml for assembly with mira.
i used the commands:
sff_extract -s sample_in.iontor.fastq -x sample_traceinfo_in.iontor.xml sample.sff

and all the time getting the error msg
usage: sff_extract [-h] [-o OUTPUT] [-c] [--min_left_clip MIN_LEFT_CLIP]
[--max_percentage MAX_PERCENT] [--version]
[input [input ...]]
sff_extract: error: unrecognized arguments: -s -x sample_traceinfo_in.iontor.xml sample.sff

any help???

[Jose Blanca]

It seems that you are using the new sff_extract included in the seq_crumbs package. This new sff_extract has a new interface and it has changed its behaviour a little. I recommend you to use this new version, just change the parameters accordingly.
The support for the xml file has been removed, because it is not required any more by MIRA. If you are sure that you need it for the latest MIRA versions just let me know and we'll do something.

[chayan]

Thank you for your suggession but I am afraid that the latest version of mira (mira-3.4.1.1) do need both the xml and fastq format for denovo assembly, so in that case i have to use sff_extract to get the out put in xml format as well. can you help me??

[Jose Blanca]

From the Mira mailing list:

On Aug 13, 2012, at 15:12 , sourakhata tiréra wrote:
> [...]
> The problem is that i trimmed my initial fastq file and some reads are missing because they come from symbiotes micro-orgamisms of the tick.
> Now mira dont work with the trimmed fastq file. Can mira work with 454 Data without xml file?
Yes, it can. "--notraceinfo" is probably what you are looking for.

> I thought about a mean to just give to mira the ID of reads that I dont want to cluster. Is it possible?
I'm not sure I got this question right. Do you want to cluster some reads but others not? If yes, what should MIRA do with the reads not-to-be-clustered?

B.

If you trim the reads I think that the xml file is irrelevant, but I would ask in the Mira mailing list just in case.

[chayan]

okk thnx for the info, i did not knw this...so only the fastq file is enough the run mira for a denovo?? if so, should i use "--notraceinfo" ?

[Jose Blanca]

You would had to trim the reads before feeding them to Mira. I guess tha the latest version do not require the notraceinfo, but I recommend to you to ask in the Mira mailing list, because I haven't used Mira for a while.

[chayan]

Dear Blanca,
I talked to the MIRA mailing list and Bastien recomends me to use the XML files as well for MIRA. He also informed that in the latest version of seq_crumbs package mate-splitting tools are also mising. Additionallly, yes MIRA can use clipped data btu for a better amd more accurate result XML is also required specially when working with Ion torrent data,
Here's why
"E.g.: assume a sequence without left clips, but with two right clips. The quality clip at 15, the adaptor clip at 52 (visualised with spaces in the next line)

GTACGATCGAAAAA aaaaaaaaaaattttttttttttttttttttaaaaaa gtgtgtgtgt

Now, Ion has sometimes interesting homopolymer artefacts (not homopolymer errors, but artefacts like above) and MIRA makes sure they get completely clipped. I.e., for MIRA the above sequence become to

GTACGATCG aaaaaaaaaaaaaaaattttttttttttttttttttaaaaaa gtgtgtgtgt

Note that one of the clips advanced by 5 bases to the left, clipping it another couple of bases and completely hiding the Ion artefact. Now, if people use clipped sequence only, all MIRA would see is:

GTACGATCGAAAAA

Note how the homopolymer artefact, which was not entirely clipped in the SFF, now still contributes with 5 A to the sequence, and here MIRA is absolutely unable to see the true nature of those (most of the time) totally wrong A bases."
this was elustrated to me by Bastien

Thanks
chayan