MindWalk's Preclinical Dengue Data Support a HYFT® Computational Prediction: Immunogens Designed on the Predicted Target Elicited Cross-Reactive Antibodies Against Antigens of All Four Dengue Serotypes, Reproduced Across Two Independent Campaigns

HYFT® technology computationally identified and prioritized a pan-serotype dengue target before immunization. Preclinical data support the prediction at the binding level: immunization with immunogens designed on the predicted target elicited antibodies that cross-react with antigens of all four dengue serotypes, and scrambled-sequence controls support that the response is specific to the designed immunogen rather than background reactivity.

MindWalk Holdings Corp. (NASDAQ:HYFT) today reported preclinical dengue data that support a computational prediction its HYFT® platform generated before any animal was immunized. HYFT® computationally identified a candidate pan-serotype structural target on the dengue virus. The subsequent immunization data support that prediction at the antibody-binding level: animals immunized with immunogens designed on the predicted target generated antibodies that cross-reacted with antigens of all four dengue serotypes — a clear and consistent trend, reproduced at the cohort level across two independent campaigns that used distinct immunogen formats and adjuvants. Scrambled-sequence controls did not show the same cross-serotype recognition, indicating the response was specific to the designed immunogen.

The result carries implications beyond the dengue program. AI-driven drug discovery depends on biological ground truth — experimentally grounded signal connecting a computational prediction to a measurable outcome in living systems. That substrate has been largely absent from the field: generated inconsistently, outside regulated frameworks, and rarely structured to meet the demands of enterprise AI deployment. The HYFT® platform is designed to produce it. Today's data are MindWalk's first public evidence that it can.

Today's data document the platform delivering on its fundamental claim: HYFT® technology computationally prioritized the pan-serotype target before immunization, the experiment supported the binding-level prediction, and the experimentally supported target now anchors MindWalk's dengue pipeline IP position as a persistent asset within the company's Biointelligence Platform. The same predictive capability applied in dengue is being deployed across MindWalk's influenza program and other announced programs where multi-strain coverage presents the analogous design challenge.

"The central question for any AI drug discovery platform is whether its computational predictions correspond to real biology," said Dr. Jennifer Bath, PhD, Chief Executive Officer and President of MindWalk Holdings Corp. "HYFT® is built to identify functional patterns that are constrained by biophysical reality — patterns that evolution cannot easily escape because they are tied to structural and binding requirements. We identified the dengue target region not because it appeared conserved at the sequence level, but because HYFT® recognized it as functionally constrained in a way that conventional sequence analysis would not prioritize.

"When animals immunized with immunogens built around that prediction produced antibodies that recognized antigens from all four dengue serotypes — and scrambled-sequence controls did not — the prediction was supported in a live immunological system. That is a closed prediction-to-result loop, and it is the first time we have documented one on this vaccine platform. The antibodies' functional profile — whether they neutralize virus and how they interact with ADE pathways — is what the next body of work will determine. What the current result tells us is that the platform identified real biology. The question now is how broadly that capability holds."

The Market: An Unsolved Pan-Serotype Problem in a High-Burden Infectious Disease

Dengue infects an estimated 400 million people annually across four antigenically distinct serotypes. Approved dengue vaccines reach only previously seropositive populations because cross-reactive antibodies against one serotype can enhance, rather than neutralize, disease upon first exposure to a second — antibody-dependent enhancement (ADE). The unsolved pan-serotype problem has kept a significant portion of the endemic seronegative population outside the addressable market for dengue vaccination, and has made a validated pan-serotype immunogen IP position the critical upstream asset for every next-generation dengue vaccine, broad dengue antibody, and pan-flaviviral platform program in development.

Identifying that position requires operating below the level of sequence homology. Serotype-to-serotype sequence identity across dengue envelope domain III (EDIII) ranges from 74 to 83 percent — high enough to generate broadly cross-reactive antibodies, insufficient to distinguish a functionally constrained pan-serotype target from a cross-reactive decoy.

HYFT® is designed to make that distinction. Rather than relying on sequence alignment or predicted immunogenicity — the parameters conventional target selection leans on — HYFT® identifies functional patterns shaped by biophysical constraints: structure, charge distribution, and binding geometry that remain conserved despite evolutionary diversification because biology cannot easily abandon them. Within MindWalk's Biointelligence Platform, these HYFT® patterns function as biological anchor points that connect sequence information to structural context, literature-supported biology, and experimental evidence. That connection is made through LensAI™, MindWalk's contextualization and reasoning layer, and the Company's Knowledge Graph — a persistent biological intelligence layer that accumulates and links evidence across programs.

The significance of the current dengue data is therefore broader than a single dengue vaccine-design experiment: it provides initial experimental evidence for a repeatable discovery loop in which computationally prioritized biological patterns are tested in vivo, and the resulting evidence can be reused across future discovery programs. The current data support cross-serotype antibody recognition at the binding level; future studies will determine whether these antibodies neutralize virus and assess their relationship to antibody-dependent enhancement.

"Animals immunized with the immunogens we designed around the HYFT®-identified target generated antibody responses that recognized antigens from all four dengue serotypes, while scrambled-control immunizations did not show the same cross-serotype recognition profile," said Dirk Van Hyfte, MD, PhD, Chief Technology Officer of MindWalk Holdings Corp.

"HYFT® computationally prioritized the target region before immunization, and the subsequent antibody-binding data are consistent with that prediction. For MindWalk, the important result is not only the observed binding response, but also the demonstration that computational biological anchors can be connected to structural analysis, immunogen design, experimental readouts, and a reusable evidence layer."

Platform Implications

MindWalk's Biointelligence Platform is built around three complementary components. HYFT® identifies biologically meaningful anchor patterns across sequence and structural space. The Knowledge Graph connects those patterns to structural, functional, literature, and experimental evidence. LensAI™ provides the contextualization and reasoning layer that enables scientists and AI models to operate over a common biological representation. Together, these components are designed to create a reusable biological intelligence infrastructure rather than a workflow optimized for any single disease program.

MindWalk's Knowledge Graph currently contains hundreds of millions of biological patterns and tens of billions of biological relationships accumulated through years of biological data integration and enrichment. Each experimentally supported result contributes additional evidence to this growing representation, strengthening its potential value across future programs in infectious disease, immunology, protein engineering, and therapeutic discovery.

The dengue study also demonstrates the platform's ability to identify both broadly conserved and biologically differentiated patterns within the same analytical framework — a capability particularly relevant to discovery programs where balancing broad coverage and selective targeting represents a central design challenge.

About MindWalk Holdings Corp.

MindWalk Holdings Corp. (NASDAQ:HYFT) is a bio-native AI company building the biological intelligence infrastructure that life sciences AI and agentic AI require. Its proprietary HYFT system functions as the AI-ready data substrate for life sciences -- a function-aware representation organized around HYFT pattern-objects that span sequence and structural biology, refined over 15 years of curation by a continuously evolving Knowledge Graph of 660 million biological patterns and 25 billion relationships, that enriches data at ingestion and compounds in analytical value with every program run on the platform. The LensAI platform operates on this infrastructure to enable target discovery, candidate diligence, and portfolio decision support, and to host the agentic AI workflows pharma is racing to deploy. MindWalk's architecture is designed so that the HYFT representation layer -- not any individual AI model -- is the durable, compounding competitive asset. The Company operates antibody discovery sites in Victoria, British Columbia and Fargo, North Dakota.

Forward-Looking Statements

This press release contains forward-looking statements within the meaning of applicable United States and Canadian securities laws. Forward-looking statements are often identified by terms such as "expects," "intends," "plans," "anticipates," "believes," "estimates," "targets," or similar expressions, or by statements that certain actions, events, or results are expected to occur or be achieved. Forwardlooking statements in this press release include, without limitation, statements regarding: the current status and anticipated advancement of the Company's wholly owned pan-serotype Dengue program, which is at the preclinical, in vivo proof-of-concept stage; the anticipated characterization of whether the cross-reactive binding antibodies described in this release neutralize virus and of their relationship to antibody-dependent enhancement (ADE), as to which no results have been disclosed; the interpretation of the preclinical binding data as supporting a computational prediction generated by the HYFT® platform before immunization; the characterization of HYFT® pattern-objects and the LensAI™ platform as capable of identifying functionally constrained biological targets and of supporting a repeatable discovery loop whose resulting evidence can be reused across future programs; the characterization of the experimentally supported dengue target as a persistent asset within the Company's Biointelligence Platform and The Knowledge Graph; the application of the same predictive capability across the MindWalk Holdings Corp. | Nasdaq: HYFT | Page 4 Company's influenza program and other announced programs where multi-strain or multi-serotype coverage presents an analogous design challenge; the characterization of The Knowledge Graph's scale (660 million biological patterns and 25 billion relationships); the anticipated provision of pipeline updates as milestones are reached; and statements regarding the structural, functional, or competitive characteristics of the HYFT® biological pattern system and the LensAI™ platform. Cross-reactive antibody binding observed in preclinical studies does not ensure that such antibodies will neutralize virus, will avoid antibody-dependent enhancement, or will translate into clinical advancement or commercial success of any program. Forward-looking statements are based on management's current expectations and assumptions and are subject to known and unknown risks, uncertainties, and other factors that may cause actual results to differ materially, including: risks relating to the outcome of ongoing and future in vivo studies, neutralization assays, and ADE characterization; the ability to advance programs from proof-of-concept to clinical development; regulatory determinations; competition; intellectual property risks; the technical performance of AI-based discovery methods; the ability to enter into pharmaceutical or other partnership arrangements; and capital markets conditions. Additional information is available in MindWalk's Annual Report on Form 20-F and other filings on SEDAR+ (sedarplus.ca) and EDGAR (sec.gov/edgar). Except as required by law, MindWalk undertakes no obligation to update any forwardlooking statement. Nothing in this press release constitutes investment advice. This press release does not constitute an offer to sell or the solicitation of an offer to buy any securities.

Trademarks

HYFT®, LensAI™, HYFT Base™, HYFT Matrix™, HYFT Prime™, and B cell Llama™ are trademarks of MindWalk Holdings Corp. or its subsidiaries. All other trademarks are the property of their respective owners.

Source: MindWalk Holdings Corp.

View source version on businesswire.com: https://www.businesswire.com/news/home/20260602948506/en/

Investor Contact

Louie Toma, CPA, CFA

Managing Director, CoreIR

investors@mindwalkai.com