Q1. We cant really find the bell shaped distributions in our samples - except for in MH0032 where there are two higher coverage distributions. Q2. This is because we have many organisms with relative low abundance. This makes it very hard to distinguish them using coverage information. Q3. It will probably perform much like SOAPdenovo. Q4. There arent really any differences in the coverage distributions between the assemblers, the metagenome assemblers are having as many problems as the standard assembler. The coverage distribution tells us that by far the most contigs have fairly low coverage in the assembly and this is also what we expect. Q5. There arent really any major differences between the assemblies. The MetaVelvet is shorter and has fewer scaffolds, but also have fewer N-bases. The main difference is probably in the number of very short contigs (<500nt). Q6. MH0032: 86755 ; MH0047: 62024. 2-4 times of human genes. Q7. We only count it once because they are from the same DNA fragment - ie. it was only present once. Q8. If each pair map to a different gene then we count it as one hit to each gene, because we have seen them both once (our DNA fragment just happened to be spanning both). Q9. There are both genes in common and genes specific for each sample. Most of the genes have very low abundance (the blue field near 0). Q10. Many of the species have very few genes, so we could probably not really trust all of them. We need to have a better reference genome set that covers more of the genomes in our samples (human gut). We could blast vs. human gut species instead. Q11.Yes there are!