The comparison of the average time spent in obtaining results from HLAMatchmaker using the conventional and automated methods revealed that the EpHLA find more software was almost 6 times faster when used by manual analysis experts (experienced group) and over 10 times faster when used by users with low analysis experience (Table 3, t-test, p < 0.0001). The class II HLA analysis required a longer average time to perform for both conventional ( Table 3; t-test, p < 0.002) and automated ( Table 3; Mann–Whitney, p < 0.0001) programs when compared to the class I HLA analysis. No difference in the number of non-self eplets was reported by users after both types of analyses: it was counted a total of 72,908 non-self

eplets in HLA class I and 58,762 non-self eplets in HLA class II. However, disagreements were observed with respect to the categorization (colors) given to some eplets between the conventional and automated methods. In fact, there was one disagreement for HLA class I and eleven disagreements for HLA class II eplets. These twelve eplets were classified as reactive (black) in the conventional analysis and as non-reactive (blue) in the automated analysis. As a consequence of such eplet categorization, twenty-one HLA alleles were considered

UMMs, when using the conventional analysis, whereas they were classified as AMMs when using the automated analysis. Due to these 21 AMMs’ disagreements, the number of HLA alleles considered AMMs in the conventional approach Meloxicam was 10,737, however check details in the automated approach 10,758

HLA alleles were considered AMMs. A closer examination of the above reported results revealed that there were errors in eplets’ categorization when using the conventional HLAMatchmaker analysis. In particular, Fig. 1 shows a case with disagreements due to human error in conventional analysis. The revised analysis permitted the correct categorization of eplets as non-reactive and the respective HLA molecules as AMMs. Fig. 1 shows screenshots of categorization eplets’ disagreements between conventional and automated HLAMatchmaker analysis. The assigned cutoff was 500, alleles in bold were assigned was AMMs. The eplets 57PS and 125SH should be blue in conventional analysis (panel 1A), because they are present on bead 47 with negative reaction of MFI = 67 as shown by automated analysis (panel 1B). Also, the allele DQB1*05:02 in conventional analysis should be in bold (panel 1A), because it is an AMM with blue non-self eplets as shown in automated analysis (panel 1B). All disagreements identified in this study occurred due to human errors made by the non-experienced group during the conventional HLAMatchmaker analysis. However, the comparison between two methods showed no statistically significant difference for these variables (class I eplets, p = 0.99; class I AMMs, p = 0.85; class II eplets, p = 0.42 and class II AMMs, p = 0.14).