4.6  Predictions of group B proteins
Two kinds of potential true negative or false negative were distinguished in the study. The group that appeared to do particularly well at predicting those proteins that do not binding sialic acid or sialic acid glycan was, perhaps not surprisingly, group B, i.e. those proteins not expected to bind any kind of sugars. Out of the original sample of 10 proteins not expected to bind any kinds of sugars significantly, and that also confirmed that expectation, i.e. true negatives as far as predicting sialic acid binding is concerned, two (i) hemoglobin and (ii) trypsin precursor for which the prediction plots for which are shown in Fig. 4. The others not shown are mostly quite large proteins containing more than 5 tryptophan residues (W) sites, for which the method still correctly predicted as non-sialic-acid binders, and so worthy of some comments. They included (iii) human ubiquitin C of 685 residues and no tryptophans (W) and no residue score exceeding 56, in contrast to (iv) human progesterone receptor of 933 residues of which 6 were tryptophan but none of which exceeded 100 (one had the highest score of 95). The remaining true negative cases are (v) fatty acid oxidation complex subunit alpha FadB of Acinetobacter calcoaceticus of a substantial 717 residues, (vi) the mitochondrial NADH-ubiquinone oxidoreductase 75 kDa subunit of the camel, which comprised a substantial 733 residues but did notably not exceed a score of 77 for any residue, (vii) human cytochrome C with a maximum score of 87 for the second tyrosine (Y) in TGQAPGYSYTATAANKN, and (viii) alcohol dehydrogenase (human, 1A) that did not exceed a score of 81 for any residue despite the “concern” that ethanol having basic sugar-like features and so could, a priori, be marginal. Perhaps unfairly included as rather small, (ix) proinsulin nonetheless does contain a tryptophan (W) which correctly did not exceed 100 and indeed only had a score of 72 in LLALLALWGPDPAAA; the phenylalanine (F) in GPDPAAAFVNQHLCG had a highest score of 81 in the sequence. In the initial study, the case most closely approaching a false positive in this group was (x) human prothrombin with a substantial number of 622 residues, there was only one residue, glycine (G), that reached a score of 100, and a residue score should exceed 100 to classify the whole domain or protein as sialic acid binding. It is possibly best declared as an example of a marginal case. At the outset false positives were expected to appear in this non-sugar-binding group groups as the sample is increased, not least because of the preliminary nature of the method. Human angiotensin converting enzyme type 2 (ACE2) was the first exception found and it is a significant exception in that it had two substantial regions 198–276 and 599–610 both exceeding scores of 100 throughout and peaking at substantial scores of 112.