Memory requirement and path for database to run metagenome-atlas

[ramnageena11]

Hi,
I have 70 Gb shotgun illumina PE data from 4 samples with 3 replicates.
I need your help regarding Running of metagenome-atlas for analysis:

How much RAM do I need? I have a workstation Dell 7920 with 96 Gb RAM with 26 core (52 cpu) and processor Intel(R) Xeon(R) Gold 6230R CPU @ 2.10GHz.
Is it enough to run the samples for downstream analysis?

I have installed the metagenome-atlas in a new environment (conda). Do i need to download the databases and link the path?
consider me beginner as suggest for setting up the tool and running it on the samples.

Thanks
rgds
RNS

[SilasK]

Probably no. I am not shure if you can do the assembly. You would pool the 3 replicats, would you? But you can always try.
Maybe with megahit.

[ramnageena11]

Hi SilasK,
Yes, I have to pool the 3 replicates for assembly.
I have done the assembly using MEGAhit. Now can I use metagenome-atlas using assembled contigs? I am planning to analyse MAGs also.

Thanks
Ram

[SilasK]

You did the assembly with megahit in atlas. Ok go ahead with atlas run binning and atlas run genomes

[ramnageena11]

Hi Silas, I did the Megahit assembly separately. Now I have megahit assembly (from 3 replicates of samples (6 files of reads)). So, can I proceed with it in metagenome-atlas? Pls suggest! Thanks rgds Ram

…

On Fri, Feb 2, 2024 at 8:28 AM Silas Kieser ***@***.***> wrote: You did the assembly with megahit in atlas. Ok go ahead with atlas run binning and atlas run genomes — Reply to this email directly, view it on GitHub <#14 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AK4LETX3B2H72HMSHSJ3CI3YRUA2ZAVCNFSM6AAAAABCAWWOJKVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTSMRUGEYDEMZWGU> . You are receiving this because you authored the thread.Message ID: ***@***.***>

[SilasK]

I am worried if you add the assembly for each sample that atlas get confused because it has the same contig multiple times.

What I suggest you is to create two atlas projects:

1. All replicates combined (Qc reads)
2. All replicates separately

Add your assemblies to booths sample.tsv.
But you need to rename the assemblies to sample_nr e.g. sample1_0, sample1_1 …
You can do this with rename.sh installed with atlas.

With project 1.
Run atlas run assembly, then binning, then genome.
Check that atlas doesn’t restart any assembly.

To get info about the differences by replicate.
in project nr2

• Run atlas run assembly. Which should give you some assembly mapping statistics.
• Add ‘genome_dir: path/to/project1/genomes/genomes’ to the config file of project 2. Then you should be able to run atlas run quantify_genomes to get abundance of the genomes per replicate.