See if this helps (http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0047656).

They seem to have used soapdenovo for assembly. I know this was used by the metahit consortium to generate their integrated gene catalog. 

Hi,

1/ First of all, what kind of metagenomics sequences do you have? How many Gb/Tb? Do you target a special specie/organism?

Take also a look to others metrics of quast like number of predict genes, number of assemblies bases etc. Take also a look to the number of assemblies reads. In fact the N50 can be significantly skewed if the assembly output contains a large number of small contigs.

2/ I don't use this, but you have "MetAMOS: a modular and open source metagenomic assembly and analysis pipeline" (http://www.genomebiology.com/content/14/1/R2) who compares multiple assemblers.

If you realy need to assemble the metagenome sequences, MetAmos works fine but spend a lot of time processing. Try MIRA 4.0, always works and if you have any question they help you. Of course, here too. 

go to galaxy server from the link bellow and you will find the tools you need

https://usegalaxy.org/

Thank you all for the inputs.  Concat tool is good with the exception that its not giving me option to choose assembler  except soapdenovo.

Hi Corentin Hochart,

Thanks for answering. I have data 50 GB/28samples. Why i said N50 is because i am dealing with bacterial metagenomes and the size varies for bacterial genomes and N50 cant be correct criteria in that case ofcourse for single genomes N50 is established. Correct me if i am wrong. Yes metmos is good one to work with. 

http://www.genomebiology.com/2015/16/1/243

The substrative assembly strategy could bring you better results.

Hi i use IDBA_UD for metagenome.

http://bioinformatics.oxfordjournals.org/content/early/2012/04/06/bioinformatics.bts174

http://i.cs.hku.hk/~alse/hkubrg/projects/idba_ud/

If you used Illumina, IDBA and especially IDBA_UD is a good tool to assemble metagenome. In fact, it works on a scale of kmers.

You'r right. N50 is a good metric to evaluate metagenomes, but ranking assemblies just by their N50 can be very misleading if the majority of the contigs are very short. That's why, I also check other metrics like gene prediction for the metagenome assembly. After it depends of your main goals!

Hope this will help you

http://www.nature.com/nbt/journal/v31/n6/full/nbt.2579.html

Omega doesn't work. Megahit is the best plus it can use multiple kmers and is highly scalable. 

Whats the sample type? How complex is it? How much RAM do you have access to?