Document status: Validation Report
Software: ABIS Online
Reference software: Apiweb™ (bioMérieux)

1. Validation Objective
The objective of this validation study was to evaluate the ability of ABIS Online to provide reliable and accurate identification of bacterial strains based on their morphological, physiological, and biochemical characteristics. The validation was conducted according to two complementary approaches:
a. Verification of identification performance: assessment of whether reference strains were correctly identified by ABIS Online and whether the software assigned a high identification probability (ID%).
b. Comparative assessment: comparison of ABIS Online identification results with those obtained using the validated commercial Apiweb™ software (bioMérieux), with particular attention to agreement in taxonomic identification and similarity of identification probabilities.
The purpose of the comparative assessment was not to establish absolute equivalence between the two software systems, but to evaluate the consistency of their identification outputs when applied to the same reference strains and biochemical data.

2. Materials and Methods
2.1 Reference Strains
A total of 50 reference bacterial strains were included in the validation study. The strains were obtained from recognized culture collections, including: ATCC, NCIMB, NCTC, DSM, CCM, local bacterial culture collections.The selected strains represented a range of bacterial genera and species with different morphological and biochemical characteristics.
2.2 Identification Procedure
The morphological and biochemical characteristics of each reference strain were evaluated using "ABIS Online*" and "Apiweb™ (bioMérieux)".Where applicable, all laboratory methods available for characterization were considered, including: classical biochemical tests, physiological tests, API bioMérieux identification strips, other routinely available phenotypic identification tests.
The results obtained from the laboratory testing were entered into both identification systems according to their respective operating principles and available test databases. For each strain, the identification provided by ABIS Online was compared with the corresponding Apiweb result.

3. Results
The comparative identification results for the 50 reference strains are presented in the accompanying validation results table.
Overall, ABIS Online demonstrated a high level of agreement with Apiweb™ for the tested reference strains. For the majority of strains, both systems produced the same taxonomic identification, with differences primarily involving the reported identification probability. ABIS Online was also able to provide an identification for several strains for which Apiweb™ either failed to provide a species-level identification or proposed a different taxon. The individual cases in which the results differed materially are discussed below.

4. Discussion
4.1 Azotobacter chroococcum was identified by ABIS Online with an identification probability of 100%. Apiweb™ did not provide a definitive identification and instead proposed Sphingomonas paucimobilis, together with four additional taxa, without assigning an identification percentage. This result demonstrates that ABIS Online was able to achieve a more specific identification for this reference strain under the conditions of the validation.

4.2 Lysinibacillus sphaericus was correctly identified by ABIS Online, whereas Apiweb™ returned the result “Bacillus non-reactive”.This outcome is consistent with the challenging biochemical profile of L. sphaericus, which does not ferment most sugars and produces numerous negative biochemical reactions. Consequently, identification systems relying on a restricted biochemical profile may have difficulty distinguishing this organism from other non-reactive Bacillus-related taxa.

4.3 ABIS Online identified Clostridium tetani with an identification probability of 44.6%. Although this probability is comparatively low, C. tetani was included among the alternative results generated by Apiweb™. Apiweb™ initially reported Clostridium spp. at 75.3% and Clostridium botulinum / sporogenes at 22.5%, followed by C. tetani as an alternative identification without an associated identification percentage. The results therefore indicate partial agreement between the systems, although the available biochemical profile did not permit a high-confidence species-level identification.

4.4 Proteus hauseri is biochemically similar to Proteus vulgaris, with differentiation depending in part on esculin and salicin reactions. These tests were not available either in the laboratory testing performed for this validation or within the applicable API galleries. The ATCC 13315 strain of P. hauseri was therefore identified by ABIS Online as “P. vulgaris / P. hauseri”, while Apiweb™ reported “P. vulgaris group.” The difference is considered consistent with the limitations imposed by the available biochemical test set and does not represent a contradictory identification.

4.5 Bacillus subtilis was correctly identified by both ABIS Online and Apiweb™. However, neither system was able to reliably differentiate the strain from closely related species with highly similar biochemical profiles, including B. atrophaeus, B. vallismortis, and B. amyloliquefaciens.
A relevant discrepancy was observed for citrate utilization. The citrate test included in the API 20E strip was negative after 48 hours of incubation, whereas the classical Simmons' citrate test was positive after 24 hours. Apiweb™ used the API 20E result, while ABIS Online incorporated the classical biochemical test result. This illustrates the importance of considering the methodology and incubation conditions associated with individual biochemical reactions when interpreting automated or database-assisted identification results.

4.6 Erysipelothrix rhusiopathiae was correctly identified by both systems. ABIS Online assigned an identification probability of 95.2%, whereas Apiweb™ did not provide an identification percentage. The API system lacked several tests relevant to the differentiation of this organism, including PYZ, alpha-GLU, beta-NAG, and GEL. This limitation prevented Apiweb™ from achieving a more precise identification. ABIS Online was not subject to the same restriction because the system permits the user to enter a broader range of available biochemical and physiological test results rather than limiting identification to a predefined test panel.

4.7 Lactobacillus reuteri cannot be reliably differentiated from Lactobacillus fermentum using a limited set of conventional physiological and biochemical tests. Definitive differentiation may require additional methods, including genotypic identification. Both systems therefore returned L. fermentum. ABIS Online additionally indicated L. reuteri as a possible alternative. The result is considered consistent with the known phenotypic similarity between these organisms and demonstrates the importance of interpreting database identification results in the context of the discriminatory power of the available tests.

4.8 ABIS Online correctly identified Lactobacillus casei with an identification probability of 98.2%. Apiweb™ identified the strain as L. paracasei subsp. paracasei, with a reported probability of 99.8%. This discrepancy represents a species-level differentiation that cannot be considered definitive based solely on the phenotypic results available in this validation. The case emphasizes the importance of considering the taxonomic resolution and database composition of individual identification systems.

4.9 Staphylococcus intermedius. The API Staph identification system does not reliably differentiate Staphylococcus intermedius from closely related staphylococci in all cases. Apiweb™ may return Staphylococcus aureus while indicating the possibility of S. intermedius, particularly when the strain is of veterinary origin. This limitation is relevant to veterinary diagnostic laboratories, where strains originating from animals may be susceptible to misinterpretation if the alternative identification is not considered. The observation illustrates that identification results should be interpreted together with the source of the isolate, clinical or veterinary context, and relevant phenotypic characteristics, rather than solely on the highest database probability.

4.10 Apiweb™ did not identify Streptococcus cristatus and instead returned Gemella morbillorum at 94.9%, together with Streptococcus mitis biovar 1. ABIS Online correctly identified S. cristatus with an identification probability of 91.2%. This result demonstrates a relevant difference in the discriminatory capacity of the two databases for this reference strain.

4.11 Streptococcus gordonii. ABIS Online identified Streptococcus gordonii with an identification probability of 79%. Apiweb™ incorrectly identified the strain as Gemella morbillorum. Although the ABIS probability was below the level generally associated with a highly confident identification, the correct species was nevertheless ranked as the principal identification by ABIS Online.

4.12 Taylorella equigenitalis was identified only by ABIS Online, with an identification probability of 46%. Both systems proposed Moraxella species among the alternative results.The relatively low identification probability indicates that the available phenotypic profile did not provide sufficient discriminatory information for a high-confidence identification. Nevertheless, ABIS Online was able to include the correct taxon among its identification results, whereas Apiweb™ did not provide the correct species as the principal identification.

5. Overall Assessment
The comparative evaluation demonstrated a high degree of concordance between ABIS Online and Apiweb™ for the reference strains included in this validation. The differences observed between the two systems can be attributed to several factors, including: differences in the underlying identification databases, differences in the number and type of biochemical tests supported, differences in the interpretation of individual biochemical reactions, phenotypic similarity between closely related bacterial species, limited discriminatory power of conventional biochemical testing for certain taxa and differences in the calculation and presentation of identification probabilities.
Importantly, several discrepancies did not represent a direct contradiction between the two systems but rather reflected differences in taxonomic resolution or the inability of one or both systems to discriminate between closely related taxa.
ABIS Online demonstrated an additional practical advantage by allowing the incorporation of a broader range of laboratory test results rather than restricting identification to a predefined commercial test panel. This feature may improve identification performance in cases where additional classical biochemical or physiological tests provide useful discriminatory information.

6. Conclusion
Based on the results obtained from the evaluation of 50 reference strains, ABIS Online demonstrated the ability to provide reliable and consistent bacterial identification results. The majority of identifications were concordant with those generated by the validated commercial Apiweb™ system, with differences generally limited to identification probability, alternative taxa, or cases involving organisms with closely related biochemical profiles.
ABIS Online correctly identified several reference strains for which Apiweb™ did not provide a definitive species-level identification, including Azotobacter chroococcum, Lysinibacillus sphaericus, Streptococcus cristatus, and Streptococcus gordonii. In other cases, both systems demonstrated limitations that were attributable to the intrinsic discriminatory power of the available biochemical tests.
Overall, the validation results support the conclusion that ABIS Online is suitable for bacterial taxonomic identification based on morphological, physiological, and biochemical characteristics, provided that identification probabilities and alternative results are interpreted in conjunction with the quality of the laboratory data and the biological context of the isolate.
The validation study therefore demonstrates that ABIS Online achieves a high level of taxonomic identification performance and produces results comparable to those obtained with Apiweb™ for the reference strains evaluated.

7. Validation Statement
On the basis of the experimental results and comparative assessment described in this report, ABIS Online is considered validated for the intended purpose of supporting bacterial identification from phenotypic and biochemical test results.
For isolates producing low identification probabilities, multiple closely related candidate taxa, or results inconsistent with the expected biological characteristics of the isolate, additional confirmatory testing should be considered. This is consistent with good microbiological practice and reflects the limitations inherent to phenotypic identification methods rather than a specific limitation of ABIS Online.