Puzzle - Strange Contig Assembly In Newbler - Trimmed Ends

Question

[Biomonika (Noolean)]

Hi guys,

I have used newbler to assembly contigs from 454 reads. When I display my .ace file in Geneious, most contigs look the way I would expect them to look like - there is a consensus sequence and many reads mapped. However, some contigs look similiarly like on the picture. I understand that the most probable explanation is, that those purple regions just do not have enough coverage to be counted as something reliable that does make consensus.

However, there are some contigs which have less coverage, but ARE in consensus.

It seems like this trimming is associated with contig quality - if most of the contig has good coverage, then the low coverage regions are considered to be rather suspicious.

But still, I would expect that there is something like "minimum coverage and its quality" and that every contig is treated the same way. I would like to ask more experienced for opinion. (I am not giving here specific parameters I used because I believe this is not parameters reliant).

Thanks.

Answer 1

[lexnederbragt]

Newbler, in contrast to assemblers before it, may place parts of reads in different contigs. In the ace file, the entire read is included for each contig, even the part not aligned to that contig, but to 'the next' one. If you check the ace file, you will most likely find another contig (perhaps contig00016) that starts with the purple parts of the reads shown for your contig00015. You could rerun the assembly using the '-rip' option, this prevents read splitting. However, you may find the assembly to be slightly less good.

If you pardon the self-plug, I am describing this in more detail in my blog: http://contig.wordpress.com/2010/02/09/how-newbler-works/