assembly validation

[meriem]

hello, i'm trying to do denovo assembly for a prokaryote strain, using different software ( Abyss, Dnastar, CLC) and i got good results, 27 contigs a hight N50, and the right size of my genome, what i don't understand is when i blast all my contigs to the reference at once they cover only 50 % or less, but when i blast them one by one it seems that they cover pretty much all the reference,
i did try mauve but it align only few of my contigs, as shown, can anyone please help me to interpret the results.
should i do the assembly again
i have also did a mapping to the reference using reads and got 85 identity

[GenoMax]

Can you say something about how big is this genome? How much theoretical fold coverage you have in your data (Illumina)? Was all of the data used for assemblies?

Have you run QUAST on your assembly?

[meriem]

thank you for your reply,
my genome is 3,6 mb, the coverage is 140.
yes i have already used quast, i think there was some fragments that were not used.

[JohnN]

How about annotating it with prokka and seeing what the ORF coverage to the DNA is via Artemis. Your genome could be diverged from the reference.

[meriem]

yes i have already annotated with quest and get the following results,
bases: 3691729
tRNA: 72
CDS: 3720
rRNA: 12
repeat_region: 1
tmRNA: 1
i didn't get what you mean by orf coverage ?

[JohnN]

If you have lots of complete ORFs or genes, with similar numbers to other members of the same species, with no gaps, you can be confident that you have a fairly completed genome considering your notsohigh coverage vs the reference.

[GenoMax]

You have not said if you tried SPAdes? From all I have seen that is probably the best aligner out for prokaryotic genomes.

[meriem]

thank you, i have already tried to run artemis but i find it hard to work with, i will try again

[meriem]

yes i tried SPAdes but i didn't get satisfying results

[colindaven]

There is a decent tool called Circoletto which might help rechecking your analysis, as it does the blast and circos visualisation itself.

http://tools.bat.infspire.org/circoletto/

[meriem]

There is a decent tool called Circoletto which might help rechecking your analysis, as it does the blast and circos visualisation itself.

http://tools.bat.infspire.org/circoletto/

does it take a lot of time to do the work, cuz i've been running it since yesterday and still not done yet

[Ali May]

For a quick and dirty microbial assembly validation, one of the most useful things I've found so far is the dev version of BRIG. If I have <15 contigs, I save each of them as separate fasta files and then use BRIG to see how they align to my reference. An example result is attached.

PS: I save my contigs as separate fasta files and then perform the alignment so that I can have an idea about their order too. If you're not interested in that you can just use a single fasta file that contains all your contigs/scaffolds.

[meriem]

For a quick and dirty microbial assembly validation, one of the most useful things I've found so far is the dev version of BRIG. If I have <15 contigs, I save each of them as separate fasta files and then use BRIG to see how they align to my reference. An example result is attached.

PS: I save my contigs as separate fasta files and then perform the alignment so that I can have an idea about their order too. If you're not interested in that you can just use a single fasta file that contains all your contigs/scaffolds.

hi ali, i tried what you said but i got a very large picture, and according to the results i hv some contig that are the same ?

[Ali May]

Well first of all congratulations on getting BRIG (dev) to work without any further instructions! Very nice picture. Secondly, indeed it looks like you have two contigs covering the same region. I wouldn't worry so much about it or hate the assembler, at the end of the day this is one the reasons we look at our assemblies, right? If you can tell us the assembler you used and the type of Illumina data you are using (read length, paired/single-end), I'm sure there are many people here who wouldn't mind thinking along with you.

One idea: in BRIG you can also plot the GC content of the reference as a ring, in addition to what you show in your picture. A possibility is that a property of this region such as the GC% confused the assembler.

hi ali, i tried what you said but i got a very large picture, and according to the results i hv some contig that are the same ?