Table 3.
Significant random forest discriminatory variables.
| Metabolite | Bin (ppm) | NPC1 vs. HC | HET vs. HC | MGS vs. HC | NPC1 vs. HET |
|---|---|---|---|---|---|
| Lipids | |||||
| -CH3 (HDL) | [0.81 .. 0.83] | ↓(3) | |||
| -CH3 (VLDL) | [0.83 .. 0.89] | ↑***(7) | ↑***(2) | ↑***(6) | |
| -(CH2-)n (HDL) | [1.21 .. 1.23] | ↑***(7) | ↓ (9) | ||
| -(CH2-)n (LDL) | [1.23 .. 1.25] | ↑***(8) | ↑***(3) | ↑***(4) | |
| -(CH2-)n (VLDL) | [1.25 .. 1.31] | ↑***(2) | ↑***(5) | ↑***(1) | ↑(8) |
| -CH 2CH2CO | [1.53 .. 1.61] | ↑***(1) | ↑**(8) | ↑***(3) | ↑(2) |
| CH2-CH 2-CH= (exclusively unsaturated lipid function) | [1.96 .. 2.05] | ↑***(5) | |||
| Unsaturated lipid >CH=CH< (bin 1) | [5.26 .. 5.32] | ↑***(3) | ↑***(1) | ↑***(2) | |
| Unsaturated lipid >CH=CH< (bin 2) | [5.32 .. 5.37] | ↑***(4) | ↑***(4) | ↑***(5) | ↑ (6) |
| Amino Acids | |||||
| Isoleucine | [0.913 .. 0.936] | ↑***(6) | ↑(4) | ||
| Proline/CH2-CH 2-CH= (bin 1) | [1.94 .. 1.965] | ↑***(9) | |||
| Proline/CH2-CH 2-CH= (bin 2) | [1.95 .. 1.96] | ↑***(6) | |||
| Histidine (bin 1) | [7.02 .. 7.08] | ↑(7) | |||
| Histidine (bin 2) | [7.74 .. 7.85] | ↑ (5) | |||
| Metal ions | |||||
| Ca2+-EDTA | [2.52 .. 2.58] | ↑***(7) | ↑*(1) | ||
| Other | |||||
| 2,3-butanediol | [1.13 .. 1.15] | ↑(8) | |||
Arrows indicate an increase/decrease in the measured metabolite with respect to the HC samples or with respect to the HET samples in the case of the NPC1 v. HET comparison. The results of the Tukey’s HSD test for each metabolite identified by the random forests analysis are indicated by an asterisks (Bonferroni corrected p-values < 0.05, 0.01, 0.001 are represented by *, **, and *** respectively). The order in which the discriminatory variables were ranked by random forest analysis based on their MDA value is represented in brackets (1 = discriminatory variable with highest MDA).