Choosing a de novo assembly for DGE analysis

Question

[User 4014]

Hi folks,

I have two assemblies that are built from k-mer 25 and 31. I ran TrinityStats.pl and busco (eudicot_odb10) to compare them (below). In general k-mer 31 provides better stats, except complete and single buscos that is better with k-mer 25. I prefer k-mer 31, but I am not so sure. May I have your opinion about which one should I go with, please? The organisms I am working with is a tree species and it is polyploid.

Thank you very much in advance!

K-mer = 25

|C:94.3%[S:36.5%,D:57.8%],F:3.3%,M:2.4%,n:2326    |
|2192   Complete BUSCOs (C)                       |
|848    Complete and single-copy BUSCOs (S)       |
|1344   Complete and duplicated BUSCOs (D)        |
|76 Fragmented BUSCOs (F)                     |
|58 Missing BUSCOs (M)                        |
|2326   Total BUSCO groups searched    

Contig N10: 4040
Contig N20: 3062
Contig N30: 2464
Contig N40: 2001
Contig N50: 1590
Median contig length: 489
Average contig: 902.20
Total assembled bases: 403475639

k-mer = 31

|C:95.2%[S:35.3%,D:59.9%],F:2.8%,M:2.0%,n:2326    |
|2215   Complete BUSCOs (C)                       |
|822    Complete and single-copy BUSCOs (S)       |
|1393   Complete and duplicated BUSCOs (D)        |
|64 Fragmented BUSCOs (F)                     |
|47 Missing BUSCOs (M)                        |
|2326   Total BUSCO groups searched         

Contig N10: 4570
Contig N20: 3532
Contig N30: 2897
Contig N40: 2385
Contig N50: 1955
Median contig length: 547
Average contig: 1049.84
Total assembled bases: 445559417

Answer 1

[xonq]

Out of these two, I would definitely choose K-mer 31. I think the decrease you're seeing in complete and single-copy BUSCOs is because more fragmented BUSCOs have been assembled into duplicate BUSCOs.