This is a test version of Biostars. For the public version, visit https://www.biostars.org.
How Can I Do Scaffolding With The Single End Data ?

hello... i did assembly of reads and i got the contigs To this i used MIRA software. Now i wanted to do the scaffolding. i thought to use SSPACE or Bambus tools. i think both needs paired end data but my data is single end data. SSPACE needs a library.txt file i created a library.txt file with only one filename that was generated from the output of MIRA assembly(unpadded.fasta file). But it did not worked. Please can you help me ? thank you.

ngs contigs scaffolding
• 8,307 views • link
updated by martenboetzer2 • martenboetzer2 • written by Arpitha Hebbar Arpitha Hebbar

4 answers

Hi, I am the developer of SSPACE and wanted to give a comment on the OP's post. You CAN make paired-end data from your single read data, but it is only valuable if you have a significant read length. Say you have 454 data, you can make a paired-end library by taking the start (read1) and end (read2) of the long read. Reverse complement read2 and you have a Illumina paired-end read-like dataset. Since you know the length of the 454 reads, you also know the insert size.

• 112 views • link
written by martenboetzer2 • martenboetzer2

+1 This is clever, I had not thought of this approach. However, I wonder what one would actually gain from doing this. My intuition is that just mapping the long reads to the contigs and/or assembling the long reads as is would be a more direct approach (i.e., leading to a more contiguous assembly though not technically scaffolds). I would also be worried about the mapping ability of these fake pairs since the fake reverse read would contain more errors.

• 173 views • link
written by SES SES

my data is generated from ion torrent PGM. what is the meaning of significant read length ? is it platform dependent? my longest read length was 127507.

• 0 views • link
written by Arpitha Hebbar Arpitha Hebbar

Significant read length means that you can generate the fake pairs from both ends of the read, and still have a gap between the reads that will be useful for scaffolding (probably want at least 2X the read length).

Is that read length a typo? If you have 127 Kb reads, why are you worried about scaffolding? :-)

• 0 views • link
written by SES SES

I think you cannot make that work, you say it yourself: both needs paired end data but my data is single end data. You cannot do scaffolding without that extra bit of information to connect contigs into scaffolds. Unfortunately, the answer is that you need to do additional sequencing to scaffold, e.g. paired-end and fosmid sequences or other large-insert clones. Traditionally, BAC-end sequences have been used to achieve sequence pairs. http://www.cbcb.umd.edu/research/assembly_primer.shtml#scaffolding

For completeness: there might be a few possible ways, but they all require additional sequences or information. These possibilities should be considered in the planning phase of the sequencing approach (optimally, at least ;) and not after finding that available data don't meet the requirements)

  • some sort of physical or linkage map to map contigs to chromosomes
  • use closely related species for a reference guided assembly (mainly for really closely related species or strains)
  • maybe one could use sequence composition criteria to assign neighboring contigs, but that wouldn't give you any distance information, I personally don't think, it is worth the effort, as getting paired-end sequences should be reasonably cheap
  • ???
• 117 views • link
written by Michael Dondrup Michael Dondrup

thank you.... is there any other tool/software for scaffolding the single end data ? or is that necessary to do the additional sequencing ? please help me ...

• 0 views • link
written by Arpitha Hebbar Arpitha Hebbar

You must do additional sequencing. There is no way software can help you without that. There is simply no information in single end sequences that allows to join multiple contigs into a scaffold.

• 0 views • link
written by Michael Dondrup Michael Dondrup

Aren't the contigs enough for your project? Perhaps the scaffolding is not necessary since you already have your contigs anyway.

• 0 views • link
written by sentausa sentausa

Good point! with you 'must' do additional sequencing I meant, in case scaffolds are really required. This depends totally on what OP wants to do with the sequences.

• 0 views • link
written by Michael Dondrup Michael Dondrup

i have four samples each having the contigs of 113,87,151,129 respectively. but i need to do the scaffolding.

• 0 views • link
written by Arpitha Hebbar Arpitha Hebbar

in any software is there any option to convert the single end data into paired end data ?

• 0 views • link
written by Arpitha Hebbar Arpitha Hebbar

I am very sorry, but to repeat myself: the information is not there, there is NO WAY. single-end reads contain no information about pairs, otherwise they would be paired-end! Thus they cannot be 'converted'. If there was a way to 'infer' pair information, it would be some sort of "magic" or "alchemy", and I don't believe in that.

• 18 views • link
written by Michael Dondrup Michael Dondrup

thank you sir for giving a valuable suggestions ...

• 0 views • link
written by Arpitha Hebbar Arpitha Hebbar

I think you are confused about what paired-end and single-end means. You can perhaps use a paired-end as single-end but the reverse cannot be so. It might not be possible to generate a whole new data by itself.

But you can try making another copy of the data, add /1 and /2 for headers as prefixed for the reads and this is to be used as paired-end, but BOTH IN THE SAME DIRECTION. It creates jsut data-duplicity which can be used as paired-end. This is IF AND ONLY IF you have no other way.

The best solution is as Michael Dondrup said, another sequencing for paired-end.

• 0 views • link
written by Rohit Rohit

"Faking" the paired-end data will still not add the information needed for scaffolding.

• 1 views • link
written by skymningen skymningen

Log in to answer this question.

More posts like this