Fig 1.
Experimental workflow for batch Y2H-NGIS. After the mating between bait and prey, diploids go through a non-selective culture to reach exponential phase. Once there, the culture is split into two flasks, one for non-selection and another for selection. After reaching saturation in each condition, culture aliquots are taken to be sequenced.
Fig 2.
Effect of count normalization in Y2H-NGIS.
A) PCA analysis of raw read counts and B) TPM, C) RUVs, D) Median-of-ratios and E) Library size normalized reads for selected (left) and non-selected (right) samples for five bait screenings (color coded). F) Boxplot of the pairwise correlation coefficients for raw and normalized read counts for all samples in non-selected and, separately, selected conditions. G) Coefficient of variation (CV) for each prey using different normalization methods in the three MLA6 baits and two luciferase screenings. Higher CV values may indicate poor performance because of a high variation between replicates.
Fig 3.
Principle of Y2H-SCORES and performance in an ideal scenario.
A) Y2H-SCORES is comprised of the enrichment score, which detects changes in prey proportions in selected and non-selected conditions; the specificity score, which measures differences in the prey enrichments with different baits under selection; and the in-frame score, which identifies the enrichment of prey reading frames under selection, assigning higher values to in-frame preys. B) to D) ROC curves of the enrichment, specificity, and in-frame scores in an ideal scenario. Colored sections represent 95% confidence intervals for the score values 0.7, 0.5, and 0.3. E) PCA of the Y2H-SCORES calculated under the ideal scenario.
Fig 4.
Effect of changes in the parameters that define Y2H-NGIS simulation.
Examples of challenging scenarios were simulated to determine the Y2H-SCORES classification power. A) Stickiness (percentage of auto-active/non-specific preys in the library), B) Strength of true interactors, C) Overdispersion, D) Concentration of true interactors in the prey library, E) Number of replicates, and F) Number of baits. Receiver Operating Characteristic (ROC) and Precision Recall (PR) AUC values were reported for the enrichment, specificity, and in-frame scores.
Fig 5.
Benchmarking of Y2H-SCORES with published Y2H-NGIS datasets.
Receiver Operating Characteristic (ROC) and Precision Recall (PR) AUC values for Y2H-SCORES, the Borda ensemble and the reference scoring method from each study.
Fig 6.
Experimental validation of the interaction between the MLA6 NLR receptor and barley targets.
A) Binary Y2H test between MLA6 and fourteen barley prey targets showing the diploid controls (SC-LW), stringent interaction (SC-LWH + 1 mM 3AT), and tests with empty-bait vector to show the specificity of the interaction. Preys are ranked in order of Y2H SCORES Borda ensemble. Rows followed by an asterisk (*) designate protein was also identified with MLA61-161 as shown in S4 Fig. Rows with two HORVU I.D.s indicate duplicate copies in the genome. B) Y2H-SCORES performance and PCA of scores calculated from the MLA6 datasets. C) Prediction of interologs for the validated Y2H-SCORES interactions with MLA6. Significant trans-eQTL associations (q-value<0.001) [39] with the Mla1 (locus coordinate 1H.05) and MlLa (locus coordinate 2H.67) are color coded. Biological processes associated with interologs are depicted in boxes next to each group.