Skip to main content
. 2021 Jul 26;22:214. doi: 10.1186/s13059-021-02419-7

Table 1.

Comparison of STRONG, DESMAN and mixtureS for strain reconstruction in the synthetic community data sets

Method Data set MAGs #SCGs #fSCGs Found Not F. Rep. Err R2 fG
STRONG Synth_S03 20 33.18 19.71 45 20 3 0.069 0.81(0.99) 26/38 = 0.68
DESMAN 42 23 2 0.125 0.81(1.00) 26/38 = 0.68
mixtureS 35 30 8 0.706 8/38 = 0.21
STRONG Synth_S05 21 32.13 22.71 53 13 2 0.062 0.87(0.99) 29/38 = 0.76
DESMAN 50 16 9 0.222 0.83(0.96) 20/38 = 0.53
mixtureS 38 28 9 0.590 7/38 = 0.18
STRONG Synth_S10 25 32.05 22.72 58 19 0 0.036 0.84(0.99) 30/43 = 0.70
DESMAN 58 21 10 0.206 0.79(0.92) 25/44 = 0.57
mixtureS 43 36 7 0.279 9/44 = 0.20
STRONG Synth_S15 23 32.17 24.19 60 12 1 0.046 0.87(0.97) 34/42 = 0.81
DESMAN 58 14 9 0.144 0.81(0.92) 25/42 = 0.60
mixtureS 38 34 16 0.321 9/33 = 0.21

Data set: Results are shown for the four different sample numbers. MAGs: The number of MAGs reconstructed with more than two reference strains. #SCGs: The average number of SCGs found in each MAG. #fSCGs The average number of SCGs after filtering in STRONG. Found: Number of strains matched to a reference strain. Not F.: Number of reference strains that had no predicted strain with a closest match to it. Rep.: Number of strains matching to a reference that already has a better match. Err: The average error rate of the ‘Found’ strains in percentage base pairs. R2: Correlation between predicted and actual strain relative proportions given as adjusted R2, the figure in parentheses is when restricted to MAGs where the number of strains was correctly predicted. fG: the fraction of MAGs where the number of strains was correctly inferred