Use of all reads from different samples to generate master assembly

[morning latte]

Dear experts,

I have seen some studies that used all reads from different samples (metagenomic datasets) to generate master assembly. Then map reads from each sample to the master assembly and calculate its contribution or recruitment rate of each sample to the master assembly. I do not fully understand main purpose of doing this. Why comparing contribution rate of reads from each sample is interesting to see? Could someone help me understand this? Any help would be greatly appreciated.

[Brian Bushnell]

It's certainly important in terms of assembly completeness. If only 20% of your reads map to your assembly, then it is not very complete.

For multiple samples, it's a bit more nuanced. If they were from, say, 3 different depths in a lake, and the top depth mapped at 90%, middle at 80%, and bottom at 40%, that tells you something. What is it? Probably depends on your study. But at a minimum, you know that if you want a good representation of deep water organisms, you should probably do something differently, like assemble them separately from the other samples or generate more sequence. And any claims you make about the composition or functions at that depth would be much less reliable than the other depths.

[morning latte]

Dear Brian Bushnell,

Thank you for your help all the time. You comment is very helpful.