Table 1.
Resulting structural model learned from a causal network
| Breed | Conditional independent |
Conditional dependent with one parent |
Conditional dependent with two parents |
Conditional dependent with multiple parents |
|---|---|---|---|---|
| LR | [MCHC] | [BAS|MON] | [HMG|MCHC:IL10] | [RBC|HMG:HMT:MCV:MCH] |
| [PLT] | [EOS|PLT] | [HMT|HMG:MCHC] | [MCV|HMG:HMT:PLT] | |
| [MON] | [IL8|TNF] | [MCH|MCV:MCHC] | [WBC|HMT:EOS:HAP:IL8] | |
| [HAP] | [IL12|IFN] | [IL6|IL10:IL1b] | [LYM|NEU:MON:EOS:BAS: TNF] | |
| [IFN] | [IL10|IFN:IL12] | [NEU|RBC:WBC:MON:BAS] | ||
| [TNF|IFN:IL10] | [IL1b|WBC:EOS:IL10:IL12] | |||
| [IL4|EOS:IL10:IL1b:TNF] | ||||
| LW | [MON] | [IL12|HAP] | [MCV|IL12:IL6] | [RBC|HMG:HMT:MCV:MCH:MCHC] |
| [HAP] | [HMG|MCH] | [HMT|HMG:MCHC] | [WBC|RBC:HAP:IL1b] | |
| [IFN] | [MCHC|MCV] | [MCH|MCV:MCHC] | [NEU|HMT:MON:HAP:IFN: IL8] | |
| [PLT|RBC:WBC] | [LYM|NEU:MON:EOS:BAS] | |||
| [BAS|WBC:NEU] | [EOS|MCV:PLT:WBC:IL8] | |||
| [IL1b|IL10:IL12] | [IL10|HAP:IFN:IL12] | |||
| [IL8|HMT:WBC] | [IL4|IL10:IL1b:IL6] | |||
| [IL6|IFN:IL10:IL1b] | ||||
| [TNF|MON:IFN:IL12:IL6] |
LR = Landrace, LW = Large White, RBC = Red blood cells, HMG = Hemoglobin, HMT = Hematocrit, MCV = Mean Corpuscular Volume, MCH = Mean Corpuscular Hemoglobin, MCHC = Mean Corpuscular Hemoglobin Concentration, PLT = Platelets, WBC = White blood cells, NEU = Neutrophils, LYM = Lymphocytes, MON = Monocytes, EOS = Eosinophils, BAS = Basophils, HAP = Haptoglobin, IFN-γ = Interferon-γ, IL = Interleukin, TNF-α = Tumor Necrosis Factor-α. Conditional dependencies are indicated as straight line. Local structure is presented in square brackets [] with the first string identifying a node. Parents of the node are listed after “|” and are separated by colons “:”. Children of the node represent the interaction determined by the conditional probability, derived from two or more parent nodes. These parental relationships are also indicated in different colors for arrows in Fig. 1. The causal network model was assigned in three categories for more comprehensive understanding of the model structure. Conditionally dependent traits identified by the network structure given in [] were used as trait combinations for multivariate genome-wide association study.