newbler 2.6 scallfold problems

[plumb]

I used newbler 2.6 with the -scaffold option and hope I can use the paired end fosmid library sequences (insert size is around 30-40kb) to produce some scaffolds. It did produced some scaffold files but no of them are really scaffolds. They are exactly the same as the contigs. In another word, all the scaffold only contain one contig.

I have total around 6753 paires in total of 13506 reads of the fosmid sanger sequences.


When I look at the 454PairStatus.txt file, all the paired sequence' status under the following situation: BothUnmapped, OneUnMapped, NoThreshold.

One I examine the NoThreshold ones, they are all mapped to different contigs.

I could not find any things which refer to NoThreshold in the manure. Could anyone give me some hints?

[flxlex]

The clue is the 'NoThreshold' status: these appear when there were not enough reads with pairs mapping to the same contig for newbler to determine the average distance between the end. You will need to have enough coverage to generate long enough contigs for at least some of the pairs to map to the same contig...

Another thing to know is that newbler requires at least two pairs lining two contigs before a join can be made.

Sanger pairs usually come in low coverage. So, another one trick to try would be to duplicate all reads, effectively feeding two copies of each into the assemlky. However, newbler reduces exact duplicates to one copy (checking for alignments that start at the exact same position and show the same sequence). So, you would have to remove a few bases from the beginning of your duplicates, e.g.

read 1         ACGTGATCGTAGGCTAGCT....
read 1_copy         ATCGTAGGCTAGCT....

If this doesn't help, try making three copies. The drawback could be that chimeric pairs at two copies or more could introduce false joins...

Hope this helps!

[plumb]

Hi Flxlex

Recently, I have found another problem with Newbler in general for the metagenome data. When doing the metagenomic data assembling, some genes cannot be assembled together. However, the targeted assembly (blast out all the reads and do the assembly with those reads), can assembly them together very nicely with the same threshold. How do you think that? what's your suggestions?