new a generic framework for fast production and evaluation of … · 2016. 1. 18. · a generic...

A generic framework for Fast production and evaluation of emergency Vaccines

A GENERIC FRAMEWORK FOR FAST PRODUCTION AND EVALUATION OF EMERGENCY VACCINES

SUMMARY: PART I OF II

Introducing the project

FastVac is a consortium of seven public health institutes and a university from seven European countries. The consortium was started in April 2010 and was co-funded by the members of the FastVac Consortium and the Health Programme of the European Union. The consortium held a final meeting in April 2014. The consortium was organized in nine work packages. The six core work packages (WP), individually and collectively aimed to accelerate key steps in the vaccine discovery and the development process. Three horizontal work packages provided the managerial support to the project.

FastVac’s remit was to produce ‘a comprehensive set of predictive rules that will enable accelerated development, evaluation, production and release of emergency vaccines’. To this end, work by the consortium was directed towards carrying out an extensive systematic literature review (SLR) of the scientific record by text mining to identify predictors of vaccine success or failure (work package 4).

The FastVac consortium has conducted two separate SLRs that focus on two parallel pathways within the vaccine development process:

• SLR1 deals with the design and testing of candidate vaccines based on probable mechanisms of protective immunity and predicted correlates of protection (COPs). Knowledge of protective immune mechanisms and identification of potential COPs is required to design and test candidate vaccines and can facilitate their licensure and release in the absence of full clinical trials.

• SLR2 deals with the design the manufacturing process and the use of appropriate process analytical technology to optimise vaccine quality and yield. These are essential requirements for rational vaccine scale-up.

Plague of Justinian> 25 milllion ? (541-542 AD)

Spanish Flu~50 million (1918-1919)

Black Death> 75 million (1340-1770)

Third PlaguePandemic

> 12 million (1855-1959)

Yersinia Pestis

Influenza A virus

Asian Flu~2 million (1957-1958)

Hong Kong Flu>1 million (1968-1969)

Russian Flu> 1 million

(1889-1890 & 1977- 978)

2009 Flu15.000

(2009-2010)

Smallpox Variola virus

> 300 million (BC-1979)

MeaslesMeasles virus> 200 million

(BC-1963) AIDSHuman immunodeficiency virus

> 25 million (1981-today)

Co-funded by the Health Programme of the European Union

Co-funded by the members of the FastVac Consortium

The FastVac Project Consortium consists of: Intravacc, NL National Institute for Public Health & the Environment, NL Public Health England, UK Statens Serum Institute, DK Cantacuzino Institute, RO National Centre for Epidemiology, HU Norwegian Institute of Public Health, NO Medical University Plovdiv, BG

www.fastvac.eu - [email protected] - June -2014

Institute for translational vaccinology - Intravaccwww.intravacc.nl

Public Health England - PHE www.hpa.org.uk

Statens Serum Institute - SSI www.ssi.dk

Cantacuzino Institute - CIwww.cantacuzino.ro

National Center for Epidemiology - NCEwww.oek.hu

Norwegian Institute of Public Health - NIPHwww.fhi.no

Medical University Plovdiv - MUPwww.meduniversity-plovdiv.bg

National Institute for Public Health and the Environment - RIVM www.rivm.nl

Experts from work package 5 to 8 assisted work package 4, in four expert groups by building and testing technical elements for the SLR, e.g. ontology development, formulation of questions, standard document collection, expert evaluation. The output of the SLR by text mining contribute to a review on correlates of protection (COPs). The principles of this review were applied to rationally design two candidate vaccines to a specific pathogen (work package 5). The candidate vaccines were tested for the induction of the relevant COPs in an animal model available within the consortium (work package 8). In work package 6 partners contibuted to the project by developing a fast track scenario for vaccine process development, scale up and technology transfer. In work package 7 partners have applied Process Analytical Technology to different vaccine platforms to aid product release.

Novel concepts will have profound regulatory implications. Work package 9 addressed some of some of the implications in a discussion paper on ‘minimum safety requirements for emergency vaccines’ (MSRV) for dialogue with regulatory agencies, industry and other relevant stakeholders. Throughout the project stakeholders were approached to start a dialogue on acceptance and implications of the project results. The literature mining tool (FastVac prospector), the main achievements and outputs of the work packages are summarised on the next page and were shared and discussed with various stakeholders during the final meeting.

1

WP 4: Technical Coordination and derivation and testing of rules

WP 9:Minimum safety sequirements for emergency vaccines & liaison for vaccine deployment

Horizontal work packages 1- 3: Coordination-dissimination-evaluation

Cor

e W

ork:

Pac

kage

s 4-

9

WP 6: Rational scale up &

process development in representative vaccine

and production platforms & technology transfer

WP 7:

PAT applied to different vaccine

platforms

WP 8:

Preclinical testing &

formulation

WP 5:

Candidate prediction and identify COPs


A GENERIC FRAMEWORK FOR FAST PRODUCTION AND EVALUATION OF EMERGENCY VACCINES

SUMMARY: PART II OF II

Main achievements and outputs and how to access them

Work package 4The key objective for FastVac work package 4 was to systematically review the literature about successful and failed vaccines and from this, make evidence-based predictions about how to optimise vaccine design and accelerate the testing, manufacture and release of vaccines.To achieve this objective, FastVac needed to review an evidence base that spans immunology, vaccinology and process technology and draws information from a wide range of viral and bacterial diseases and across a selection of vaccine platforms. It was estimated that over 10 million documents would need to be reviewed which could not be done manually using conventional systematic review approaches. FastVac therefore had to develop a novel approach. The chosen approach uses automated text-mining to carry out high-throughput processing and annotation of documents and a bespoke user interface, the FastVac Prospector, to retrieve and analyse relevant documents; together they constitute the FastVac vaccinology decision support platform.

• The FastVac ontology will be made available as an open access resource and will be published

• Worked examples using the FastVac Prospector and FastVac Prospector SLR platforms will be published. The systems are available for evaluation on request: [email protected]

• The annotated SRL1 and SLR2 Mímir indices may be available on request: [email protected]

Work package 5Work package 5 partners have contributed in building and testing technical elements necessary to perform systematic literature reviews (SLR) based on automated text mining. Work package 5 has demonstrated that the established text mining approach can be used to generate systematic knowledge about protective antigens and immune responses (including COPs) necessary for rational vaccine design. By using the automated SLR approach, supported by manual review processes, work package 5 has deduced vaccine candidates for two selected model organisms: influenza and coxiella. Two principally different experimental influenza vaccines were successfully manufactured (work package 5) and tested in animal models for immunogenicity and protective efficacy (work package 8). The work in work package 5 has demonstrated that the established text-mining system used in combination with FastVac Prospector can serve as a valuable decision support platform for efficient evidence-based design of vaccines against a variety of emerging diseases.

Plague of Justinian> 25 milllion �? (541-542 AD)

Spanish Flu~50 million � (1918-1919)

Black Death> 75 million �(1340-1770)


> 12 million � (1855-1959)

Yersinia Pestis

Influenza A virus

Asian Flu~2 million � (1957-1958)

Hong Kong Flu>1 million �(1968-1969)

Russian Flu> 1 million �

(1889-1890 & 1977- 978)

2009 Flu15.000 �

(2009-2010)


> 300 million �(BC-1979)

MeaslesMeasles virus> 200 million �


> 25 million �(1981-today)

1 Co-funded by the Health Programme of the European Union




Work package 6Work package 6 partners have developed a fast track scenario for vaccine process development, scale up and technology transfer. Regarding process development, an optimized process for the production of influenza vaccine on eggs has been developed. Thereafter the process was successfully transferred to another partner, where the process was implemented at a larger scale without operational deviations and the product met the pre-set specifications. Next to this example for influenza vaccine production, in addition a fast track scenario was developed for optimisation of medium for bacterial growth, based on the principles of Design of Experiment.

Work package 7Work package 7 partners have applied Process Analytical Technology to different vaccine platforms to aid product release and showed that better product monitoring of vaccines can support faster process development as well as faster release of intermediate products. For example, T(EM) and DLS coupled to (A)SEC were developed to characterise influenza virus and NIR was used to analyse liposomes containing CAF09 adjuvant. Also, in-process monitoring of antigen production was shown to indeed support a faster release of intermediate product.

Work package 8Work package 8 partners have tested work package 5 candi-date influenza vaccines and optimised delivery route and/or formulation of vaccines against varicella zoster and tuberculo-sis. Furthermore, based on the automated SLR and existing consortium data on interspecies extrapolation, work package 8 partners have found that animals can indeed be used as models for vaccine efficacy when substancial knowledge is available about the disease patterns in the these models to be certain that correlations with humans are valid. A text-mining based SLR approach is thus a valuable tool to scan vast amounts of literature for such correlations and for identifying suitable animal models and saving valuable time in the initial phase of vaccine development which will even be more revelant in case of new pandemics.


WORK PACKAGE 4 TECHICAL COORDINATION AND

DERIVATION AND TESTING OF RULES: PART I OF IV

Summary

The key objective for FastVac work package 4 was to systematically review the literature about successful and failed vaccines and from this, make evidence-based predictions about how to optimise vaccine design and accelerate the testing, manufacture and release of vaccines.

To achieve this objective, FastVac needed to review an evidence base that spans immunology, vaccinology and process technology and draws information from a wide range of viral and bacterial diseases and across a selection of vaccine platforms. It was estimated that over 10 million documents would need to be reviewed which could not be done manually using conventional systematic review approaches. FastVac therefore had to develop a novel approach. The chosen approach uses automated text-mining to carry out high-throughput processing and annotation of documents and a bespoke user interface, the FastVac Prospector, to retrieve and analyse relevant documents; together they constitute the FastVac vaccinology decision support platform.

The FastVac vaccinology platform uses semantic and relational annotations rather than single keywords to retrieve, in a single search, relevant documents from a vast amount of literature that covers 128 viral and bacterial pathogens of public health significance (the ‘FastVac pathogens’) and includes a wide range of vaccine platforms, from classical inactivated pathogen approaches to more experimental systems such as synthetic peptide vaccines. In the retrieved documents, the association between different sets of variables, e.g. vaccine platforms and the type of immune response they elicit, can be visualised using FastVac Prospector and the documents containing these associations can be reviewed by the user. This enables trends and relationships to be identified, for example what type of vaccine induces cell mediated immunity and whether it does so for a particular pathogen. In this way the FastVac vaccinology decision support platform enables the scientist to make evidence-based decisions to guide the rational design, testing and manufacture of vaccines against any of the FastVac pathogens.





> 12 million � (1855-1959)

Yersinia Pestis

Influenza A virus




(1889-1890 & 1977- 978)

2009 Flu15.000 �

(2009-2010)


> 300 million �(BC-1979)




The FastVac reviews

FastVac has conducted 2 separate reviews (SLRs) that focus on 2 parallel pathways within the vaccine development process: SLR1 deals with the design and testing of candidate vaccines based on probable mechanisms of protective immunity and predicted correlates of protection (COPs). Knowledge of protective immune mechanisms and identification of potential COPs is required to design and test candidate vaccines and can facilitate their licensure and release in the absence of full clinical trials.SLR2 deals with the design the manufacturing process and the use of appropriate process analytical technology to optimise vaccine quality and yield. These are essential requirements for rational vaccine scale-up.

Process overview

The process followed for each review is summarised in figure 1. A large corpus of documents (abstracts) was extracted from online databases (PubMed, the Cochrane library and IFI CLAIMS Global Patent database) using very broad search terms. These documents were processed and annotated using a purpose built text-mining application that uses the open source General Architecture for Text Engineering (GATE) platform. The resulting index of annotated documents was loaded into the FastVac Prospector interface and queries were used to retrieve sets of relevant documents. Within the retrieved documents, associations and trends were identified using visualisation tools built into FastVac Prospector.





Figure 1: Simplified process overview for the development and use of the FastVac vaccinology decision support platform. White squares represent inputs; grey squares are process steps. Document standards were used during the development phase to evaluate the performance of the annotation and document retrieval processes. Data visualisation was performed on the full annotated document corpus (SLR1: ~9.5 x 106 documents; SLR2: ~2.3 x 106 documents) once the development phase had been completed.



DERIVATION AND TESTING OF RULES: PART II OF IV

Key components of the FastVac vaccinology platform

It has taken approximately 3 years of effort to develop and evaluate the process described above, first for SLR1 and then for SLR2. Key components that had to be developed and refined included:

The FastVac OntologyThe key terminology relating to vaccinology and immunology that would need to be annotated in the document corpus was collected into a purpose-built ontology of structured terms (entities). The FastVac ontology encompasses terms about pathogens, diseases, immunology and vaccine manufacture (Table 1). At its most complex, parts of the FastVac Ontology contain nine hierarchical levels of terminology, allowing either very broad or extremely specific analyses of the data to be undertaken.





> 12 million � (1855-1959)

Yersinia Pestis

Influenza A virus




(1889-1890 & 1977- 978)

2009 Flu15.000 �

(2009-2010)


> 300 million �(BC-1979)




FastVac text-mining applicationA text mining application was developed that splits each document in the corpus into its structural components (sentences, words, punctuation etc) and annotates (1) entities according to the FastVac ontology, and (2) key relations e.g. ‘protective’ according to bespoke term lists. This process produces an index of annotated documents.

FastVac questions and query algorithmsQuestions were designed to direct the retrieval of relevant documents from the annotated index. SLR1: 15 questions (Q1-15; Table 2) addressed key decision points along the vaccine design and testing pathway (Figure 2); Q18 & Q19 were designed to identify suitable adjuvants for particular pathogens / vaccine platforms (Table 2). Beneath each tier 1 question, 2 further tiers of ‘child’ questions were designed, of increasing specificity (not shown).





Table 2: FastVac Questions 1 For the FastVac pathogens, what antigens are protective or associated with disease

amelioration or control of infection? 2 For the FastVac pathogens, what antigens are known not to be protective or show

no association with disease amelioration or control of infection? 3 For the FastVac pathogens, what antigens cause disease exacerbation or elicit an

adverse immune response? 4 For the FastVac pathogens, what immune responses are associated with protection,

disease amelioration or control of infection? 5 For the FastVac pathogens, what immune responses are known not to be

associated with protection, disease amelioration or control of infection? 6 Is immune-mediated disease exacerbation or an adverse immune response

associated with any of the FastVac pathogens? 7 What are the known correlates of protection for the FastVac pathogens? 8 What studies/assays/tests are used to measure immune protection/COPs for the

FastVac pathogens? 9 What vaccines protect against the FastVac pathogens or the diseases they cause? 10 What vaccines do not protect against the FastVac pathogens or the diseases they

cause? 11 What vaccines elicit a humoral immune response against the FastVac pathogens or

the diseases they cause? 12 What vaccines elicit a cell-mediated immune response against the FastVac

pathogens or the diseases they cause? 13 For the FastVac pathogens or the diseases they cause, what vaccines are

reactogenic or associated with disease exacerbation or an adverse response? 14 For the FastVac pathogens or the diseases they cause, what animals or animal

modles have been / are used to test vaccines or immunisation protocols? 15 For the FastVac pathogens or the diseases they cause, what are the features of the

disease pathogenesis process? 18 For vaccines against the FastVac pathogens or the diseases they cause, what

adjuvants enhance the immunogenicity/immune response or the protective efficacy of the vaccine?

19 For vaccines against the FastVac pathogens or the diseases they cause, what adjuvants do not enhance the immunogenicity /immune response or the protective efficacy of the vaccine?

Table 1: The top two hierarchical levels of terms in the FastVac ontology.Top class Class Top class Class

Active immunity Artificial active immunity Infection Experimental infectionNatural active immunity Natural infection

Adverse response Adverse event Outcome Adjuvant outcomeAdverse immune response Disease outcomeImmunotoxicity Immune outcome

Animals and models Animal model Treatment outcomeMan Vaccine outcomeOther animal Passive immunity Artificial passive immunity

Antigen Antigen type Natural passive immunityBacterial antigen Pathogen BacteriaViral antigen Virus

Disease Bacterial disease Process and testing Analytical methodViral disease Biopharmaceutical

Disease pathogenesis Affected organ Manufacturing processClinical signs and markers Product qualityDisease course Product yieldImmunopathology Protection Correlate of protectionSample type Cross protectionTissue damage Studies assays tests In vitro immunoassayTissue inflammation In vivo studyTissue pain In vivo test

Immune response Vaccine AdjuvantAntigen presentation Development statusCell death Licence statusCytokine and receptor Vaccine platformInnate immune response Vaccine purpose

Immunisation protocol Immunisation device Vaccine valencyImmunisation route Vaccine vectorImmunisation schedule

definingprotectiveantigens

definingimmune

mechanisms

definingCOPs

definingvaccineplatform

defininganimalmodels

use COPsfor

pre-clinicaltesting

Figure 2: The 5 key decision points in the vaccine design and testing process targeted by the FastVac SLR1 questions (in blue)

1 2 4 53

Q1-Q2-Q3 Q4-Q5-Q6 Q7-Q8 Q9-Q10-Q13(Q11-Q12)

Q14-15 Testing

Adaptive immune response



DERIVATION AND TESTING OF RULES: PART III OF IV

For SLR2, 3 questions were designed to direct the retrieval of relevant documents from the annotated index. Beneath each tier 1 question, 1 or 2 further tiers of ‘child’ questions were designed (Table 3). For all questions, corresponding query algorithms (Mímir queries) were developed to retrieve documents which contained the requisite combination of entities and relations to answer each FastVac question.

FastVac ProspectorThe index of annotated documents and the query algorithms were incorporated into FastVac Prospector, designed for FastVac by the University of Sheffield. This provides a password-protected online user interface which allows researchers to retrieve relevant documents from the indexed corpus and create visual representations of the results to help them make evidence-based decisions. For a set of retrieved documents, FastVac Prospector allows particular sets of annotated terms to be selected. Associations between terms in different sets can then be visualised using one of 2 views; the association matrix or the association chart (Figure 3). To compare many terms the association matrix is most suitable.





> 12 million � (1855-1959)

Yersinia Pestis

Influenza A virus




(1889-1890 & 1977- 978)

2009 Flu15.000 �

(2009-2010)


> 300 million �(BC-1979)




Figure 3: Animal models used to test successful DNA vaccines against Mycobacteriaceae. Two-dimensional association matrix (A) produced using FastVac Prospector to analyse the documents retrieved by Q9 indicating numbers of PubMed abstracts in which types of vaccine platform (x axis) co-occur with named bacterial pathogens (y axis; grouped in to families for visualisation). Colour intensity of association matrix cells uses a relative colour scale based on the number of documents used to create the matrix. Documents selected for this visualisation were required to mention a vaccine against a relevant pathogen which produced a protective effect. Documents in which ‘Mycobacteriaceae’ co-occurred with ‘Nucleic acid vaccine’ where used in an association chart (B), showing the different species used in those studies. The width of each arc gives a relative indication of how many documents contain both terms joined by that arc.





A

B

Table 3: The tier 1 and tier 2 questions designed for FastVac SLR2. For Q1 and Q2, additional tier 3 questions divide the results into vaccines and biopharmaceuticals (not shown). No. Tier 1 Question No. Tier 2 Question 1 For vaccines and other

biopharmaceuticals, what analytical methods are used to determine product quality?

1.1 For vaccines and other biopharmaceuticals, what functional analytical methods are used to determine product quality?

1.2 For vaccines and other biopharmaceuticals, what non-functional analytical methods are used to determine product quality?

2 For vaccines and other biopharmaceuticals, what manufacturing processes influence product quality?

2.1 For vaccines and other biopharmaceuticals, what manufacturing processes influence product purity?

2.2 For vaccines and other biopharmaceuticals, what manufacturing processes influence product stability?

2.3 For vaccines and other biopharmaceuticals, what manufacturing processes influence product potency?

3 For vaccines and other biopharmaceuticals, what manufacturing processes influence product yield?

3.1 For vaccines, what manufacturing processes influence product yield?

3.2 For other biopharmaceuticals, what manufacturing processes influence product yield?

AlcaligenaceaeAnaplasmataceae

BacillaceaeBartonellaceae

BrucellaceaeBurkholderiaceae

CampylobacteriaceaeChlamydiaceae

ClostridiaceaeCorynebacteriaceae

CoxiellaceaeEnterobacteriaceae

FrancisellaceaeHelicobacteriaceae

LegionellaceaeLeptospiraceae

ListeriaceaeMycobacteriaceae

MycoplasmataceaeNeisseriaceae

PasteurellaceaeRickettsiaceae

SpirochaetaceaeStaphylococcaceae

StreptococcaceaeVibrionaceae

Ace

llula

r_va

ccin

eC

apso

mer

e_va

ccin

eC

onju

gate

_vac

cine

Ect

osom

e_va

ccin

eE

xoso

me_

vacc

ine

Fusi

on_p

rote

in_v

acci

neLi

pid_

vacc

ine

Mic

rove

sicl

e_va

ccin

eN

ucle

ic_a

cid_

vacc

ine

Spl

it_vi

rus_

vacc

ine

Sub

unit_

vacc

ine

Sub

vira

l_pa

rticl

e_va

ccin

eS

ynth

etic

_pep

tide_

vacc

ine

Toxo

id_v

acci

neVe

ctor

ed_v

acci

neVi

rus-

like_

parti

cle_

vacc

ine

Who

le_o

rgan

ism

_vac

cine

Ferre

t

Bird

Non-

hum

an p

rimat

e



DERIVATION AND TESTING OF RULES: PART IV OF IV

Performance of the FastVac system

The FastVac text mining system was developed and refined through 2-3 iterations (Figure 1). Each iteration used a new set of evaluation standards selected by domain experts to positively answer the top-level text mining questions (positive standards) or mark them to be irrelevant (negative standards). Documents that were classified inappropriately by the system were examined manually to check for missing or erroneous annotations or defects in the query algorithms. After each evaluation round the text mining application, queries and ontology were improved.The performance metrics used to evaluate the system were the standard text mining metrics of precision and recall. For SLR1, qualitative analysis by a panel of experts was also undertaken in the third round of evaluation.Question performance was optimised to prioritise high precision over high recall, in order to reduce the number of false positives and so minimise the ‘noise’ when the results were displayed visually.Final performance scores for each review, averaged across all the top-level questions were:• SLR1: 70% precision, 63% recall• SLR2: 72% precision, 70% recall

Proof-of-concept and beta testing

FastVac has produced a dynamic and interactive decision support platform for developing new vaccines. It can be used to make decisions for any of the 128 annotated ‘FastVac pathogens’ or the genera and families to which they belong.As proof of concept, FastVac has used the platform to identify mechanisms of protection, define theoretical vaccine candidates and predict potential correlates of protection for use in preclinical testing for selected exemplar pathogens. In brief:

• For model pathogens (N. meningitidis and hepatitis B virus) where the protective immune response has been well-characterised, the FastVac platform correctly identified the protective mechanism.





> 12 million � (1855-1959)

Yersinia Pestis

Influenza A virus




(1889-1890 & 1977- 978)

2009 Flu15.000 �

(2009-2010)


> 300 million �(BC-1979)




• For other pathogens (influenza A virus and Coxiella sp), potential COPs identified using the platform concurred with those identified by experts or through a conventional manual literature review.

• For a theoretical newly-emerged pathogen (a hantavirus), predictions were made by comparison with other genus and family members about the most appropriate vaccine platform, likely correlates of protection and the most appropriate animal models for preclinical testing.

To carry out a more extensive test of the FastVac approach was beyound the scope of the project. It is recommended that the tool should undergo wider beta testing by groups of subject experts, subject to further funding.

Applications

The FastVac vaccinology decision support platform shows promise for the following applications:

• For ‘dynamic’ querying in the event of an outbreak of a re- or newly-emerged pathogen e.g. by public health institutes, health organisations, SMEs and vaccine manufacturers.

• As an interface with the vaccine R&D for anyone embarking on developing a new vaccine against an existing pathogen e.g. universities, SMEs etc.

• To identify gaps in the knowledge base for research funders and vaccine developers (pathogens for which neither successful nor failed vaccines exist or for which there is no data on protective or non-protective immune responses).

• For regulators to identify if the vaccine development pathway taken by a vaccine developer is appropriate.

The outputs of FastVac and how to access them

• The FastVac ontology will be made available as an open access resource and will be published

• Worked examples using the FastVac Prospector and FastVac Prospector SLR platforms will be published. The systems are available for evaluation on request: [email protected]

• The annotated SRL1 and SLR2 Mímir indices may be available on request: [email protected]






WORK PACKAGE 5CANDIDATE PREDICTION AND COPS IDENIFICATION: PART I OF II

WP5 has contributed to the FastVac project by:

• Building and testing technical elements necessary to perform systematic literature reviews (SLR) based on text mining

• Demonstration that automated SLR can be used to generate and integrate key knowledge within vaccinology

• Designing model vaccines for influenza and coxiella based on automated SLR processes

• Manufacturing influenza model vaccines for testing in animal models

• Demonstration that SLR can be used as a decision support platform for fast and evidence based vaccine design

Establisment of automated Systematic Literature Reviews (SLR) for evidence based vaccine development:

Workpackage 5 partners have been involved in expert groups tasked to specify the pathogens studied by FastVac, provide expert advice about correlates of protection (COPs), and collect animal model data. Most effort has been devoted to assist work package 4 in building the necessary ontology elements (immunology domains) for SLR, constructing query questions used in the text mining processes to address work package 5 milestones, and evaluating the performance of the text mining by reviewing documents in a web-based expert evaluation tool (EET). Work package 5 addressed the milestones and deliverables by analyzing query questions 1 to 13 in the Fastvac Prospector (Figure 1).





> 12 million � (1855-1959)

Yersinia Pestis

Influenza A virus




(1889-1890 & 1977- 978)

2009 Flu15.000 �

(2009-2010)


> 300 million �(BC-1979)




Figure 2: The 2D maps show the Fastvac prospector results for A) Q9 (Tier 1: What vaccines protect against the FastVac pathogens or the diseases they cause?): co-cocurrence of influenza A types vs vaccine platform. B) Q1 (Tier 1: For the FastVac pathogens, what antigens are protective or associated with disease amelioration or control of infection?): co-cocurrence of influenza A types vs type of viral antigen.






definingimmune

mechanisms

definingCOPs



use COPsfor

pre-clinicaltesting

Figure 1: Vaccine development pathway: The 5 key decision points in the vaccine design and testing process targeted by the FastVac SLR1 questions (in blue). Work package 5 used Q1 - Q13 of the FastVac propector tool for analyses.

1 2 4 53

Q1-Q2-Q3 Q4-Q5-Q6 Q7-Q8 Q9-Q10-Q13 Q14-15 Testing

Acellular_vac.Capsomere_vac.

Chimeric_capsomere_vac.Conjugate_vac.Ectosome_vac.Exosome_vac.

Fusion_protein_vac.Lipid_vac.

Microvesicle_vac.Outer_membrane_vesicle_vac.Detergent-extracted_OMV_vac.

Native_OMV_vac.Nucleic_acid_vac.

DNA_vac.BAC_vac.

Polynucleotide_vac.RNA_vac.

Self-replicating_RNA_vac.Split_virus_vac.

Subunit_vac.Polysaccharide_subunit_vac.

Capsular_polysaccharide_subunit_vac.Protein_subunit_vac.

Purified_protein_subunit_vac.Recombinant_subunit_vac.

Subviral_particle_vac.Synthetic_peptide_vac.

Lipopeptide_vac.Long_overlapping_peptide_vac.

Synthetic_long_peptide_vac.Synthetic_polypeptide_vac.

TLR_ligand-peptide_conjugate_vac.Toxoid_vac.

Vectored_vac.Bacterial-vectored_vac.

Viral-vectored_vac.Amplicon_vectored_vac.

Attenuated_viral-vectored_vac.Chimeric_viral_vectored_vac.

Replication-defective_viral-vectored_vac.Replicon_vectored_vac.

Alphavirus_replicon_vac.Semliki_forest_virus_replicon_vac.

Ven._Equine_Encephalitis_virus_replicon_vac.sindbis_virus_replicon_vac.

Flavivirus_replicon_vac.Kunjin_virus_replicon_vac.

Tick-borne_encephalitis_virus_replicon_vac.Single-cycle_viral_vectored_vac.

Virus-like_particle_vac.Chimeric_virus-like_particle_vac.

Hybrid_Ty-VLP_vac.Recombinant_virus-like_particle_vac.

Virosome_vac.Whole_organism_vac.

Inactivated_vac.Beta-propiolactone-inactivated_vac.

Formalin-inactivated_vac.Heat-inactivated_vac.Live_attenuated_vac.

Live_attenuated_bacterial_vac.Live_attenuated_viral_vac.

Replication-competent_viral_vac.Replication-defective_viral_vac.

Single-cycle_viral_vac.single_cycle_flavivirus_vac.

H1_

viru

sH

2_vi

rus

H3_

viru

sH

5_vi

rus

H7_

viru

sIn

fluen

za_A

_viru

s

Viral_nonstructural_proteinPolymerase

DNA_polymeraseRNA_polymeraseAssembly_protein

Regulatory_proteinViral_protease

Viral_structural_proteinNucleoprotein

viral_membrane_proteinViral_glycoproteinEnvelope_protein

Spike_proteinCapsid_proteinMatrix_protein

Tail_proteinTegument_protein

H1_

viru

sH

2_vi

rus

H3_

viru

sH

5_vi

rus

H7_

viru

sIn

fluen

za_A

_viru

sBA






> 12 million � (1855-1959)

Yersinia Pestis

Influenza A virus




(1889-1890 & 1977- 978)

2009 Flu15.000 �

(2009-2010)


> 300 million �(BC-1979)




Manufacturing influenza model vaccines for testing in animal models:

Inactivated whole virus vaccine was manufactured by using a hen egg derived virus source. Embryonated eggs were inoculated with the H3N2 strain HK-X31 and incubated for 3 days before the allantoic fluid was harvested. Virus was purified before inactivation with beta-propiolactone (BPL). HI titer, total protein, and ovalbumin content were determined according to standard procedures, and inactivation was verified by using MDCK cells. Inactivated virus was formulated with CAF01 or CAF09 as adjuvant. Subunit polypeptide vaccine was designed based on conserved CD4+ and CD8+ T cell epitopes from the H1N1 and H5H1 viral proteins M1, NP, and PB1 combined with B cell epitopes (M2e and HA2) directly coupled to a universal T helper cell epitope sequence (PADRE). Synthetic peptides were prepared by Fmoc-chemistry and checked for purity and integrity using UPLC-MS and MT mass spectrometry. Before use, peptides were mixed with IFA and the TLR2 ligand Pam3CysSK4 as adjuvant components.

WP5: Output and main achievements:

Work package 5 partners have contributed in building and testing technical elements necessary to perform systematic literature reviews (SLR) based on automated text mining. Work package 5 has demonstrated that the established text mining approach can be used to generate systematic knowledge about protective antigens and immune responses (including COPs) necessary for rational vaccine design. By using the automated SLR approach, supported by manual review processes, work package 5 has deduced vaccine candidates for two selected model organisms: influenza and coxiella. Two principally different experimental influenza vaccines were successfully manufactured (work package 5) and tested in animal models for immunogenicity and protective efficacy (work package 8). The work in work package 5 has demonstrated that the established text-mining system used in combination with FastVac Prospector can serve as a valuable decision support platform for efficient evidence-based design of vaccines against a variety of emerging diseases.Preparing for automated SLR:

Ontology Query questions Performance testing

Identification of protective antigens and immune responses

Designing model vaccines

Influenza Coxiella burnetii

Manufacturing influenza modelvaccines:

Testing in animalmodels (WP8)

WORK PACKAGE 5CANDIDATE PREDICTION AND COPS IDENIFICATION: PART II OF II

SLR: Identification of protective antigens and immune responses for a spectrum of pathogens:

Based on the established text mining approach, work package 5 has carried out a systematic literature search to identify protective antigens, protective immune responses, and COPs for a range of viral and bacterial pathogens. Application of FastVac Prospector was used as a critical tool for visualising and analysing the results (Figure 2), enabling any single or group of FastVac pathogens to be selected for detailed study with respect to a variety of vaccinology related issues. This knowledge provided the basis for COP based vaccine design. Worked examples of principally different viral and bacterial pathogens have been presented in the project.

SLR: Designing of model vaccines based on COPs

Influenza and coxiella were chosen as two model pathogens for designing candidate vaccines according to COP based principles as deduced from literature review processes. Influenza was subsequently selected as the model disease for which two experimental vaccines were designed, manufactured at laboratory scale, and tested in pre-clinical studies (WP8). An SLR based prediction of model vaccines was performed by integrating information retrieved from the whole range of work package 5 related query questions involving protective antigens, protective immune responses, COPs, and vaccine platforms. By specifically focusing on the linkage between identified COPs and the known vaccine concepts that induce protective immune responses, relevant sets of criteria were developed for design of optimal influenza and coxiella vaccine candidates.Using these criteria, the following model vaccines for influenza were proposed for production and pre-clinical investigation in work package 8: Inactivated whole virus vaccines formulated with CAF adjuvants and polypeptide subunit vaccines based on conserved T cell epitopes (NP, M, PB) and B cell epitopes (M2e, HA2) formulated with IFA and a TLR2 ligand as adjuvant. The predicted bacterial vaccine types for Coxiella burnetii should contain extracted phase I whole cell vaccine antigens (depleted of reactogenic components) or subunit platforms with defined protein antigens like porin (29.5 kDa), OMP (27 kDa), Hsp B (GroEL), and SecB adjuvanted with detoxified phase I LPS.






WORK PACKAGE 6PROCESS DEVELOPMENT, SCALE UP AND TECHNOLOGY TRANSFER FOR VACCINES

Process development: the product

To describe fast track scenario’s for vaccine process development the target product profile needs to be well described. Within Fastvac this target is a prophylactic vaccine product against infectious diseases. To evoke a long lasting, broad immune response, modified pathogen like structures have been proven useful. In complexity and size these vaccine products are positioned between the small (nm), well-characterized, recombinant protein biopharmaceuticals and the large (µm) autologous cell-based regenerative medicine products.Recent guidelines (ICH Q11) give the choice to follow a “traditional route” for vaccine development or an “enhanced approach”, based upon scientific understanding of product and process. The traditional route is more empirical but still will have to yield a description of the process within certain specifications with a rationale that explains why these settings are chosen.

Process Development SchemeVaccine development from idea to market launch on average requires 12-15 years development time. In early stages of development some product is already required for research and it is important that the chosen expression system for this initial product is selected on the basis of a target product profile to prevent that an expression system is selected for immediate use that cannot support a robust large-scale production process later on.

Secondly, it is important to note that although during the clinical phases the development work declines, the scale-up and validation work increases. Since these process optimization activities are expensive it is of high importance whether the “freeze” of the production process is performed before or after clinical phase I/II studies. Mostly, for well-characterized vaccines scale-up and validation activities can be planned after clinical phase I/II studies, even though product comparability





> 12 million � (1855-1959)

Yersinia Pestis

Influenza A virus




(1889-1890 & 1977- 978)

2009 Flu15.000 �

(2009-2010)


> 300 million �(BC-1979)




has to be demonstrated to support process changes. For less defined products, first a small clinical phase I/II study should be planned after initial preclinical development, and this will need to be repeated after completion of the scale-up and process validation studies.

Methods supporting fast vaccine process development

There are a number of enabling technologies that can support faster process development for vaccines. These include the use of platform technology, high-throughput technology, and the use of disposables. In addition, the application of a Quality-by-Design approach, as feasible for well characterized vaccine products, may support faster development but will certainly provide better control.

Scale up and Process Validation

The scale up principles for vaccines are not different from general scale up strategies for biopharmaceuticals. The used strategies include I. avoiding real scale up by applying multiplication of small units, II. Rules of thumb for specific process steps, as maintaining column height for chromatography, III. Scale down based upon analysis from a large scale system used for a comparable product, and truly as last option IV. the iterative empirical route.

Fastvac experimental work

As part of the experimental work for work package 6 a revised process for the production of influenza vaccine on eggs has been developed at Intravacc and transferred to Cantacuzino, where the process was implemented at a larger scale. It was shown to run without operational deviations and importantly, the product met the pre-set specifications, as required for cGMP productions.





Productdevelopment

small scale process

developmentClinical

phase I / II

Scale up &

process validation

Productregistration

Clinicalphase III

New vaccine idea

Proof of PrincipleAnimal model

Proof of PrincipleHuman

Market production

Process validationProcess development

cGMP cGMP


WORK PACKAGE 7PROCESS ANALYTICAL TECHNOLOGY APPLIED TO DIFERENT

VACCINE PLATFORMS TO AID PRODUCT RELEASE

Good Manufacturing Practice requires predictability: in a set of documents a production process is described that, if executed within the pre-set specifications, will yield a product that meets the pre-set criteria for the product, ensuring safety and efficacy. Pharmaceutical products need to be approved for the market by Regulatory Authorities, such as EMA and FDA. For this, it is required to have control systems in place to enable a transpar-ent production performance evaluation. Thus, process develop-ment and production runs are primarily governed by product quality properties and product yield is of secondary importance.

In the last decades many recombinant biopharmaceuticals have been launched on the market and regulatory require-ments have evolved concomitantly. A Quality-by-Design (QbD) approach was described, starting with the description of the target product. Using assays that are appropriate for demon-strating the required quality characteristics of the target product, the production process is monitored. If this QbD is combined with on-line monitoring of the product quality, enabling faster release of intermediate product for the next process step, the approach also fits with Process Analytical Technology. With this development strategy, based upon scientific understanding, an optimal control – or predictability – is obtained for product and process.

Importantly, vaccines differ from biopharmaceuticals in several aspects, leading to the following challenges for the application of PAT to vaccine development:

1) Most vaccines are not well-characterized products and therefore it is difficult to apply QbD or PAT to the full extend. The knowledge on pathways involved in host-pathogen interaction has been increasing recently. However, practical applications of this knowledge has been proven difficult since currently it is not understood to which extent different pathways influence each other. In addition genetic and environ-mental factors will influence those interactions and thus the in vivo outcome.

2) The interpretation of PAT has inflated from on-line monitoring of product quality during the different steps of the production process to the on-line, or fast off-line, measurement of any process parameter. For many vaccines, batch release is based on potency assays that are often not robust, making it more difficult to correlate indirect signals to product potency.





> 12 million (1855-1959)

Yersinia Pestis

Influenza A virus




(1889-1890 & 1977- 978)

2009 Flu15.000

(2009-2010)


> 300 million (BC-1979)




For Work package 7 it was evaluated if better product monitor-ing of vaccines can support faster process development as well as faster release of intermediate product.Experiments performed within Work package 7 of the Fastvac project has been aimed at illustrating that it is worthwhile to increase monitoring of product quality properties to support vaccine process development. This was ranging from charac-terizing influenza virus with (T)EM and DLS coupled to (A)SEC to analyzing CAF09 adjuvant containing liposomes with NIR.

The complex less-defined products that do not fully qualify for an approach based upon scientific understanding of product and process can benefit most if data that is collected during development – of both product and process – can later be correlated to specific effects in humans. To prevent, in the future, that for less defined vaccine products it cannot be evaluated whether a declining protection is caused by a change of the circulating pathogen or a change in the manufacturing process of the vaccine product, a database of product properties and process settings might prove its value.Therefore, the recommendation of the work package 7 partners is to encourage more detailed monitoring of vaccine products and processes and not to forget to connect the data with later developed assays and results from clinical studies.





Figure 2: Comparative representation of electron-microscopy of formaldehyde inactivated influenza viruses (right) and beta-propiolactone inactivated influenza viruses (left).

Figure 1: ASEC result: beta-propiolactone versus formaldehyde inactivated whole virus bulk.


WORK PACKAGE 8PRE-CLINICAL TESTING AND FORMULATION: PART I OF III

WP8 has contributed to the FastVac project by:

• Building and testing technical elements necessary to perform systematic literature reviews (SLR) based on text mining.

• Demonstration that automated SLR can be used to generate and integrate key knowledge within vacci-nology.

• Optimizing delivery route and formulation for preclini-cal influenza vaccines designed by work package 5.

• Demonstrating that correlates of protection (COPs) defined by automated SLR can accelerate vaccine testing.

• Demonstrating extrapolation of results between species.

Establisment of automated Systematic Literature Reviews (SLR) for evidence based vaccine development:

Work package 8 partners have been involved in expert groups tasked to specify the pathogens studied by FastVac, provide expert advice and collect animal model data. Most effort has been devoted to assisting work package 4 in building the necessary ontology elements (adjuvant and immunology domains) for SLR1, constructing query questions used in the text mining processes, and evaluating the performance of the text mining by reviewing documents in a web-based expert evaluation tool (EET) as well as providing standard documents to test the performance of SLR1. Work package 8 addressed the milestones and deliverables by analyzing query questions 1 to 13 in the Fastvac Prospector and tested the use of COPs preclinically (Figure 1).

Choice of animal models for vaccine testing and optimiza-tion of route and formulation

The work package 5 candidate influenza vaccines were tested in mice and ferrets which were predicted to be good animal models for influenza by the SLR (Figure 2). The two vaccines were a universal poly-peptide and a whole-inactivated virus (WIV).





> 12 million (1855-1959)

Yersinia Pestis

Influenza A virus




(1889-1890 & 1977- 978)

2009 Flu15.000

(2009-2010)


> 300 million (BC-1979)




Figure 2: Choice of best animal model. (A) Association map generated in FastVac Prospector showing co-occurrences of documents describing different animal models (incl. humans) vs affected part of the respiratory system. The association was done on Pubmed articles on selected animals and models used to investigate disease pathogenesis for all FastVac pathogens (Q15). Scale: white (no documents) to red (maximum number of documents). (B) Association chart showing the relative contribution of articles related to disease pathogenesis which describe influenza A (red) and either of the animal models: ferrets (blue), mice (orange) and/or guinea pigs (green).1 2

They were administered to the animals by the intramuscular or intranasal route. Furthermore, the vaccines were formulated with selected adjuvants: Cationic Adjuvant Formulation (CAF) 01, CAF09, Incomplete Freund’s Adjuvant and/or Pam3CysSK4. For the WIV, the CAF09 adjuvant enhanced the efficacy of the vaccine leading to dose sparing and a broader immune response.






definingimmune

mechanismsdefining

COPs



use COPsfor

pre-clinicaltesting

Figure 1: Vaccine development pathway: The 5 key decision points in the vaccine design and testing process targeted by the FastVac SLR1 questions (in blue). Work package 8 used Q14 - Q15 of the FastVac propector tool for analyses and tested COPs (WP5) pre-clinically.

1 2 4 53


A

B

Low

er_r

espi

rato

ry_t

ract

Alv

eolu

sB

ronc

hiol

eB

ronc

hus

Lung

Ple

ural

_cav

ityP

leur

al_m

embr

ane

Trac

hea

Upp

er_r

espi

rato

ry_t

ract

Lary

nxG

lotti

sla

rynx

Pha

rynx

Lary

ngop

hary

nxN

asop

hary

nxO

rthop

hary

nxph

aryn

xN

ose

Par

anas

al_s

inus

Turb

inat

es

Animal_modelHuman_study

In_vitro_modelEx_vivo_model

ManOther_animal

ferret

Guinea_pigHamster

MouseRat

chinchillacotton_rat

deer_mousegerbil

ground_squirrelmultimammate_mouse

woodchuck


WORK PACKAGE 8PRE-CLINICAL TESTING AND FORMULATION: PART II OF III

Use of COPs to accelerate preclinical vaccine testing

Assays for the COPs identified by the SLR (work package 5) were used to evaluate the candidate influlenza vaccines (Figure 3). Thus, after vaccination, IgG, IgA, hemagglutination inhibition antibodies, CD4+ and CD8+ T-cell responses were analyzed by ELISA and flow cytometry in serum and blood samples. After two vaccinations the animals were challenged with influenza virus to address vaccine-induced protection. Most of the selected COPs correlated well with protection but interspecies correlations were hampered by the lack of reagents and assays available for the ferret model. The results support the principle of using an SLR-based approach for defining COPs to be used for preclinical evaluation of vaccines. A wider range of assays and pathogens is available in FastVac prospector (Figure 4 and 5).

Extrapolation between species

In case of a pandemic with limited time to run large clinical trials on the new vaccines it is imperative to know if results from animal studies can be extrapolated to humans. Thus, we have used new and existing data within the FastVac consortium to investigate if extrapolation between animal species can be demonstrated. Our data show that extrapolation is possible for a set of parameters but not for all. A major problem when attempting to extrapolate between species is the lack of comparable assays and reagents, e.g. in one study the most relevant biomarkers for protection against TB as determined in a human study, could not be addressed in the animal models.





> 12 million � (1855-1959)

Yersinia Pestis

Influenza A virus




(1889-1890 & 1977- 978)

2009 Flu15.000 �

(2009-2010)


> 300 million �(BC-1979)




Fastvac prospector: question 7 and 8: COPs predicted by work pakage 5 HAI antibodies - VN Antibodies - ELISA - CD4 T cells - CD8 T cells - Th1/Th2 cytokines

Work package 5 Work package 8


definingimmune

mechanisms

definingCOPs



use COPsfor

pre-clinicaltesting

Figure 3: Vaccine development pathway: The 5 key decision points in the vaccine design and testing process targeted by the FastVac SLR1 questions (in blue). Work package 8 used Q14 - Q15 of the FastVac propector tool for analyses and tested COPs (WP5) pre-clinically.

12 4 53






Thus, to be able to achieve better extrapolation between species, more has to be invested in the development of reagents and assays for all relevant animal models to make direct comparisons possible.

WP8: Output and main achievements

Work package 8 partners have tested work package 5 candidate influenza vaccines and optimized delivery route and/or formulation of vaccines against varicella zoster and tuberculosis. Furthermore, based on the automated SLR and existing consortium data on interspecies extrapolation, work package 8 partners have found that animals can indeed be used as models for vaccine efficacy when substancial knowledge is available about the disease patterns in the these models to be certain that correlations with humans are valid. A text-mining based SLR approach is thus a valuable tool to scan vast amounts of literature for such correlations and for identifying suitable animal models and saving valuable time in the initial phase of vaccine development which will even be more revelant in case of new pandemics.

Figure 4. Shows the relative contribution of documents which describe Orthomoxyvi-ruses (blue) and haemagglutination assay (purple), ELISA (red), T-cell assays (green) or neutralization assays (orange). The figure shows the Association Chart of the results presented in Figure 5 (next page).

Mouse Ferret

Man

Route of administration Non-

immunized i.m. – i.m. i.m. – i.n. Was

outcome predicted? Parameter WIV WIV+CAF01 WIV+CAF09 WIV WIV+CAF01 WIV+CAF09

Protection (weight d3) - - ++ +++ (+) ++ +++ N/A

HAI titer - - +++ +++ - + + yes IgG - + +++ +++ + ++ ++ yes IgA - - - - + ++ +++ yes Th1>Th2 - - + + - + +++ yes CD4 T cells ICS - - - - ++ +++ partly CD4 8 cells ICS - - - - - - - no CD4 8 cells dextramer

- +++ - - +++ -

- no






> 12 million � (1855-1959)

Yersinia Pestis

Influenza A virus




(1889-1890 & 1977- 978)

2009 Flu15.000 �

(2009-2010)


> 300 million �(BC-1979)




Figure 5: A wider range of assays are available for testing of the protective efficacy of vaccines against viral pathogens for mice than for ferrets or humans. Association map showing co-occurrences for the indicated species and virus documents within Fastvac prospector query Q8.3 (man: What studies/assays/tests are used to measure immune protection/COPs in response to vaccination in Man?) and Q8.4 (mouse and ferret: What studies/assays/tests are used to measure immune protection/COPs in response to vaccination in animals?). Horizontal axis = assays, vertical axis = virus families. White = no documents, red = maximum number of documents for the indicated species. The total number of documents retrieved for viral pathogens is indicated in brackets for each species.

WORK PACKAGE 8PRE-CLINICAL TESTING AND FORMULATION: PART III OF III





Man

(618

)Fe

rret

(7

9)M

ouse

(2

266)

Agg

lutin

atio

n_as

say

Bac

teria

l_ag

glut

inat

ion_

assa

yH

aem

aggl

utin

atio

n_as

say

late

x_ag

glut

inat

ion_

assa

ym

icro

scop

ic_a

gglu

tinat

ion_

test

Ant

ibod

y_m

edia

ted_

killi

ng_a

ssay

antib

ody_

med

iate

d_cy

toto

xici

ty_a

ssay

Bio

sens

or_a

ssay

Sur

face

_pla

smon

_res

onan

ce_a

ssay

Com

plem

ent_

fixat

ion_

assa

yC

ompl

emen

t_m

edia

ted_

killi

ng_a

ssay

bact

eric

idal

_ass

ayD

egra

nula

tion_

assa

yC

D10

7_as

say

gran

ule_

enzy

me_

exoc

ytos

is_a

ssay

gran

zym

e_B

_ass

aype

rforin

_ass

ayE

ffect

or_B

_cel

l_as

say

B_c

ell_

ELI

SP

OT

plaq

ue-fo

rmin

g_ce

ll_as

say

Effe

ctor

_T_c

ell_

assa

yC

ytok

ine_

secr

etio

n_as

say

CB

A_a

ssay

inte

rfero

n-ga

mm

a_re

leas

e_as

say

ELI

SP

OT

Intra

cellu

lar_

cyto

kine

_sta

inin

gcy

toto

xic_

T_ly

mph

ocyt

e_as

say

leuc

ocyt

e_m

igra

tion_

test

lym

phoc

yte_

stim

ulat

ion_

test

Enz

yme_

imm

unoa

ssay

ELI

SA

Indi

rect

_ELI

SA

chao

tropi

c_E

LIS

Aco

mpe

titiv

e_E

LIS

Agl

ycop

rote

in_E

LIS

Asa

ndw

ich_

ELI

SA

Hae

mag

glut

inat

ion_

inhi

bitio

n_as

say

Hae

mol

ysis

_in_

gel_

assa

yIm

mun

oblo

tIm

mun

ochr

omat

ogra

phy

Imm

unof

luor

omet

ric_a

ssay

Dire

ct_i

mm

unof

luor

esce

nce_

assa

yFl

ow_c

ytom

etry

_ass

ayFl

uore

scen

ce-a

ctiv

ated

_cel

l_so

rting

Pol

yspe

ctra

l_flo

w_c

ytom

etry

Indi

rect

_im

mun

oflu

ores

cenc

e_as

say

Imm

unoh

isto

chem

istry

Imm

unoc

ytoc

hem

istry

Mac

roph

age_

opso

no-a

dher

ence

_ass

ayN

eutra

lizat

ion_

assa

yTo

xin_

neut

raliz

atio

n_as

say

Viru

s_ne

utra

lizat

ion_

assa

yC

PE

_inh

ibiti

on_a

ssay

Rap

id_f

luor

esce

nt_f

ocus

_inh

ibiti

on_t

est

mic

rone

utra

lizat

ion_

assa

ypl

aque

-red

uctio

n_ne

utra

lisat

ion_

test

sync

ytia

_inh

ibiti

on_a

ssay

Ops

onop

hago

cyto

sis_

assa

yFl

uore

scen

ce_o

pson

opha

gocy

tosi

s_as

say

Kill

ing_

opso

noph

agoc

ytos

is_a

ssay

Mul

tiple

x_op

sono

phag

ocyt

osis

_ass

ayO

xida

tive_

burs

t_as

say

mac

roph

age_

oxid

ativ

e_bu

rst_

assa

yne

utro

phil_

oxid

ativ

e_bu

rst_

assa

yP

reci

pita

tion_

test

Floc

cula

tion_

assa

yIm

mun

odiff

usio

n_as

say

doub

le_i

mm

unod

iffus

ion_

assa

ysi

ngle

_rad

ial_

imm

unod

iffus

ion_

assa

yIm

mun

oele

ctro

phor

esis

coun

ter_

curr

ent_

imm

unoe

lect

roph

ores

isro

cket

_im

mun

oele

ctro

phor

esis

Rad

ioim

mun

oass

ayra

dioa

ctiv

e_an

tigen

-bin

ding

_ass

ayS

ingl

e_ra

dial

_hae

mol

ysis

_ass

ay

AdenoviridaeArenaviridaeBunyaviridae

CoronaviridaeFiloviridae

FlaviviridaeHepadnaviridae

HepeviridaeHerpesviridiae

OrthomyxoviridaePapillomaviridaeParamyxoviridae

ParvoviridaePicornaviridae

PoxviridaeReoviridae

RetroviridaeRhabdoviridae

Togaviridae









Togaviridae









Togaviridae