3 questions inss

RESEARCH ARTICLE

The journey to establishing an IT-

infrastructure within the German Biobank

Alliance

Christina SchüttlerID 1*, Hans-Ulrich Prokosch1, Michael Hummel2, Martin Lablans3,4,

Björn Kroll5‡, Cäcilia Engels2‡, on behalf of the German Biobank Alliance IT development

team¶

1 Medical Informatics, Friedrich-Alexander-Universität Erlangen-Nürnberg, Erlangen, Germany, 2 German

Biobank Node, Charité –Universitätsmedizin Berlin, Berlin, Germany, 3 Federated Information Systems,

German Cancer Research Center, Heidelberg, Germany, 4 University Medical Center Mannheim,

Mannheim, Germany, 5 IT Center for Clinical Research, University of Lübeck, Lübeck, Germany

‡ These authors are joint senior authors on this work.

¶ Membership of the German Biobank Alliance IT development team is provided in the Acknowledgments.

* [email protected]

Abstract

Background

Biobanks ensure the long-term storage and accessibility of biospecimens and correspond-

ing data sets. Thus, they form the foundation for many research projects which may contrib-

ute to improving medical care. With the establishment of the German Biobank Node and

Alliance, expertise in biobanking is bundled and strengthened. An important component

within this research infrastructure is the set-up of an information technology (IT) network for

allowing feasibility requests across individual biobanks.

Objective

We aim to describe relevant aspects that have shaped the journey to interconnect biobanks,

to enhance their visibility within the research-community, to harmonize data, and to enable

feasibility searches to support access to available data and biosamples.

Methods

To achieve this task, we resorted to a wide variety of methods: we ran a requirement analy-

sis, decided on the mode of operation for the federated team of IT-developers and on the

development approach itself, took related national and international initiatives into account,

and concluded with evaluations of the developed software artefacts and the operation of the

entire chain of applications.

Results

We drew an IT framework including all heterogeneous data aspects derived from our

requirement analysis and developed a comprehensive IT infrastructure. The successful

implementation benefited from a smooth interaction of a federated IT team distributed

PLOS ONE

PLOS ONE | https://doi.org/10.1371/journal.pone.0257632 September 22, 2021 1 / 15

a1111111111

a1111111111

a1111111111

a1111111111

a1111111111

OPEN ACCESS

Citation: Schüttler C, Prokosch H-U, Hummel M,

Lablans M, Kroll B, Engels C, et al. (2021) The

journey to establishing an IT-infrastructure within

the German Biobank Alliance. PLoS ONE 16(9):

e0257632. https://doi.org/10.1371/journal.

pone.0257632

Editor: Zeeshan Ahmed, Rutgers Robert Wood

Johnson Medical School, UNITED STATES

Received: June 4, 2021

Accepted: September 4, 2021

Published: September 22, 2021

Copyright: © 2021 Schüttler et al. This is an open

access article distributed under the terms of the

Creative Commons Attribution License, which

permits unrestricted use, distribution, and

reproduction in any medium, provided the original

author and source are credited.

Data Availability Statement: All relevant data are

within the manuscript and its Supporting

Information files.

Funding: The German Federal Ministry of

Education (https://www.bmbf.de/) funded the

German Biobank Alliance project (grant numbers

01EY1701-14). Accordingly, this work was

supported by this grant (CS and HUP: 01EY1714;

ML: 01EY1713; MH and CE: 01EY1701; BK:

01EY1710). The funder had no role in study

across all participating sites that was even able to manage a major technology change mid-

project. Authentication and project management services from associated partners could be

integrated and the graphic user interface for an intuitive search tool for biospecimens was

designed iteratively. The developed code is open source to ensure sustainability and the

local implementation is concluded and functioning. The evaluation of the components was

positive.

Conclusions

The entire project had given ample opportunity for challenges, predictable and unpredict-

able—from the mode of operation to changing some of the initial ideas. We learned our les-

sons concerning personnel, budget planning and technical as well as manual monitoring as

well as some requirements arising only during the process of the project. Nevertheless, we

can here report a success story of a network infrastructure, highly agile and much easier in

local installation than initially anticipated.

Introduction

Biobanks are an indispensable instrument of disease- and patient-oriented biomedical

research. They ensure the long-term storage and accessibility of biosamples and the corre-

sponding data sets. Thus, biobanks provide an important basis for conducting research which

boost the development of future diagnostics and precision therapy [1, 2]. Personalized preci-

sion medicine largely relies on large-scale high-throughput analyses of high-quality and well-

described disease-specific patient samples and holds great opportunities to advance and

improve modern approaches to healthcare [3, 4]. Biobanks are hubs for the acquisition, pro-

cessing, and storage of biospecimens, providing both high quality tissue and fluid samples and

access to services that process them into derivatives such as serum, plasma, DNA, and RNA.

This, in combination with linked disease and socio-demographic data of the donor, provides a

rich source for the research community [5].

In an effort to make biosamples and associated data accessible, the German Federal Minis-

try of Education and Research funded the establishment of a central organization for coordina-

tion and steering national biobank activities, the German Biobank Node (GBN) [6–8], which

also acts as a hub for the European biobank infrastructure BBMRI-ERIC (Biobanking and Bio-

molecular Resources Research Infrastructure–European Research Infrastructure Consortium)

[9]. One central pillar for this next generation biobanking is the construction of a sophisticated

information technology (IT) network. This is a crucial factor to enable higher visibility and

efficient access to biosamples and data in order to compile large, multicentric sample collec-

tions for research at national and international level. To this end, the German Biobank Alli-

ance (GBA) was formed under the umbrella of GBN.

Since the local IT situation for both, biobanking IT and the general hospital information

systems, is heterogeneous throughout Germany, the primary goal of GBA was the gradual IT

connection of biobanks to a central network infrastructure by providing appropriate interfaces

and locally missing components. The infrastructure is designed to cover two basic aspects. On

the one hand, a legally compliant IT framework had to be created to facilitate the allocation of

biosamples and associated data via a central IT platform (feasibility queries). On the other

hand, this particular platform had to be developed.

PLOS ONE Journey to establishing an IT-infrastructure for a biobank network

PLOS ONE | https://doi.org/10.1371/journal.pone.0257632 September 22, 2021 2 / 15

design, data collection and analysis, decision to

publish, or preparation of the manuscript.

Competing interests: The authors have declared

that no competing interests exist.

Abbreviations: AAI, Authentication and

Authorization Infrastructure; ABIDE_MI, Aligning

Biobanking and DIC Efficiently; BBMRI-ERIC,

Biobanking and Biomolecular Resources Research

Infrastructure–European Research Infrastructure

Consortium; BMBF, Bundesministerium für Bildung

und Forschung (Federal Ministry of Education and

Research); CDA, Clinical Document Architectur;

CQL, Clinical Quality Language; DKTK, Deutsches

Konsortium für Translationale Krebsforschung

(German Cancer Consortium); DWH, Data

Warehouse; EHR, electronic health record; ETL,

extract, transform, load; FHIR, Fast Healthcare

Interoperability Resources; GBA, German Biobank

Alliance; GBN, German Biobank Node; GUI,

graphical user interface; HL7, Health Level 7; ICD,

International Classification of Diseases; ID, identity;

IT, information technology; MIABIS, Minimum

Information About Biobank data Sharing; MII,

Medical Informatics Initiative; SPREC, Standard

Preanalytical Code.

The objective of this paper is to outline our approach of establishing and implementing this

IT infrastructure. Special attention will be paid to the exceptional situation that the develop-

ment team was scattered over several locations in Germany. Furthermore, it discusses the chal-

lenges, the lessons learned, but also the success stories brought up during this process. Thus,

we are confident that we have gained valuable knowledge in the course of this experience,

which can be of great help for similar projects.

Methods

Requirement analyses

Before starting the actual development process, the first step was to assess the initial IT situa-

tion at the local biobanks in order to identify the existing implementation gaps and to deter-

mine the demand for the planned IT solutions. This was achieved by consulting all biobank IT

managers at the participating sites. The biobank IT managers compiled and reported back the

relevant information from their site, which was then analyzed by the team. In addition, tar-

geted stakeholder workshops and surveys were held over the course of the project with IT

experts, biobank managers, scientists and patient representatives in order to gain a holistic

view. In accordance with the evaluation of these activities, the requirements were defined, reg-

ularly revised and adapted to the actual needs of the biobanks and stakeholders.

Mode of operation

Two teams were assembled to handle the required IT tasks. One team, consisting of 9 members

across 6 partner sites, was responsible for development and maintenance of the IT solutions.

The second team, consisting of one IT employee per biobank location, took care of the local

integration of the distributed IT components including their connection to the internal data

source systems and day-to-day operation. According to the agreement of the project partners,

the teams had to be organized to cooperate efficiently across many locations (Fig 1) and to

ensure close communication. Therefore, regular conference calls and face-to-face meetings

within and between the two teams took place. Additionally, a chat platform was employed for

asynchronous exchange. The development team oriented itself towards the agile scrum frame-

work [10]. This was realized by doing sprints in 2–4 weekly intervals, with planning and review

taking place at the beginning and end. In addition, short daily web conferences were held for

tight coordination. This approach supported the team in working iteratively and responding

quickly to changes in requirements.

Development approach

A strategic idea at the beginning of the project was to build on already existing IT solutions

developed in previous projects of the developers from GBA. This refers in particular to the

development of the local components. For these, the comprehensive “bridgehead” architecture

of the German Cancer Consortium (Deutsches Konsortium für Translationale Krebsforschung

[DKTK]) was used as a central element [11]. With regard to the technologies to be used to fur-

ther develop the existing components, those already established in DKTK were initially

retained for GBA in order to keep the migration effort low. Using an iterative approach, the

source code was gradually refactored to ensure maintainability and to consider modern archi-

tectural patterns. Other components and their libraries had to be replaced completely. How-

ever, the knowledge gained by using legacy components largely supported the implementation

of the replacing source code. Vice versa, enhancements were created in a manner allowing

PLOS ONE Journey to establishing an IT-infrastructure for a biobank network

PLOS ONE | https://doi.org/10.1371/journal.pone.0257632 September 22, 2021 3 / 15

them to be backported to the DKTK implementation. For example, the installation process of

the software had been modernized by GBA and could be used by DKTK.

Evaluation of components

In the course of the iterative approach ensuring that the developed components are ultimately

used according to their purpose, systematic evaluations were carried out. Two usability analy-

ses were conducted to evaluate the user-friendliness of the feasibility tool. The first one was

based on a prototype of the user interface in the development phase, while the second one was

carried out in a comparative analysis of three query builders in productive use. For both, user-

friendliness was assessed using the widely used and standardized System Usability Scale [12]

and qualitative questionnaires [13, 14]. Moreover, the functionality of an additionally con-

nected donor portal was evaluated by providing the biobanks with a demonstrator, which was

then assessed using an exemplary user scenario and an associated checklist.

Fig 1. Overview of the partners sites including biobanks, IT teams and GBN at project launch 2017. Printed under a CC BY

license, with permission from the German Biobank Node, original copyright 2017.

https://doi.org/10.1371/journal.pone.0257632.g001

PLOS ONE Journey to establishing an IT-infrastructure for a biobank network

PLOS ONE | https://doi.org/10.1371/journal.pone.0257632 September 22, 2021 4 / 15

Coordination with other initiatives

In all efforts with focus on national biobanking, attention was paid to ensuring cooperation

and coordination with similar initiatives at national as well as international level to establish

interoperability and prevent parallel structures. Therefore, a close cooperation with BBMRI-E-

RIC, as European biobanking organization, took place. The cooperation was carried out within

the context of joint working groups. The same principle was followed when working with ini-

tiatives that emerged during this project and pursued similar goals. This applies in particular

to the German medical informatics initiative (MII), which aims to enhance the secondary use

of data generated in the course of patient care [15].

Ethics statement

Approval was obtained from the responsible ethics committee before the start of each evalua-

tion study. For the first study of the Charité –Universitätsmedizin Berlin [EA1/098/19] and for

the second study of the Technical University of Dresden (Germany) was obtained [SR-EK-

262062020]. The prerequisite for participation in the studies was written informed consent

from the participants in both cases. In addition, written consent was always obtained from

external persons if they participated in workshops as part of the project. In the case of surveys

conducted among GBA project partners, consent was waived.

Results

The following section describes the IT specific content of the GBA project.

Prerequisites at the sites

The analysis of the local IT situation confirmed that some areas had not yet been covered by all

participating biobanks, in particular consent management (missing in 6 out of 12 biobanks),

project management (4/12), contact management (11/12) and donor empowerment (10/12).

Furthermore, it was necessary to elaborate a central data protection concept concerning pro-

cesses and components for the cross-linkage of biobanks that complements existing local data

protection approvals. However, there was no need for a central identity (ID) management and

pseudonymization service, since all GBA biobanks already maintained a system compliant to

the respective requirements. A further insight from the gap analysis was the use of different IT

tools in biobanks. Therefore, we aimed at a high degree of interoperability, clear definition of

interfaces and standard IT techniques.

IT infrastructure

The IT support for biosample and data search is built on a trinity. This comprises (1) the Sam-

ple Locator as graphical user interface (GUI) of a central feasibility query, (2) the local data

warehouse (DWH), and (3) the middleware as a link between biobanks and the query tool

(compare Fig 2).

The first point of contact for researchers looking for biosamples is the Sample Locator as a

web-based application for feasibility queries [16]. This enables immediate feedback regarding

the number of biosamples or donors for research projects potentially available in the biobanks.

A special feature here is the two-stage process that emerged as a requirement from the stake-

holder survey of biobank managers. This process involves a first step providing a cumulative

highly aggregated number of all biosamples fulfilling the search criteria that can be identified

across the biobanks in real-time without authentication. In the second step, registered and

authenticated requesters receive a breakdown of the aggregated biosample numbers across the

PLOS ONE Journey to establishing an IT-infrastructure for a biobank network

PLOS ONE | https://doi.org/10.1371/journal.pone.0257632 September 22, 2021 5 / 15

respective biobanks [17]. To this end, the biobanks have to install two local components into

their infrastructure specifically developed for this purpose. The DWH stores the necessary

data after harmonizing by means of an ETL (extract, transform, load) process from heteroge-

neous sample and clinical data sources using a consolidated data set. At this point, GBA could

make use of existing terminologies and standards, such as MIABIS (Minimum Information

About Biobank data Sharing) [18] and SPREC (Standard PREanalytical Code) [19]. Based on

these, a corresponding core data set for donor (basic clinical) data, sample data, and disease

specific (oncological and cardiological) data was developed in several rounds of balloting and

agreed upon in close collaboration with all GBA biobanks.

The middleware serves as an interface between DWH and the feasibility tool and transmits

the relevant information according to the request that was submitted. As this connector com-

ponent is used in several projects, a substantial work of coordination was necessary to adopt

the component successfully.

The working modus allowed flexible design decisions even during the project. For instance,

for the local DWH, we initially started with the DKTK DWH data model. However, after care-

ful examination, it turned out that its performance was too slow to allow real-time queries and

furthermore, the proprietary data format was not interoperable. Moreover, not all research

requests based on biobank inquiries could be answered satisfactorily by the underlying query

language. This insight resulted from an analysis of concrete search queries we collected from

researchers and biobanks, which we then validated against our current development status.

Since HL71FHIR1 (Health Level Seven Fast Healthcare Interoperability Resources) [20]

gained impact as a health care data transfer standard during the course of the project, the first

Fig 2. Overview of the IT infrastructure and distribution of responsibilities among the IT teams.

https://doi.org/10.1371/journal.pone.0257632.g002

PLOS ONE Journey to establishing an IT-infrastructure for a biobank network

PLOS ONE | https://doi.org/10.1371/journal.pone.0257632 September 22, 2021 6 / 15

attempt to solve this issue resulted in the switch to the FHIR1 standard query mechanism

(FHIR1 Search). This way we could overcome the lack of interoperability. However, FHIR1

is more intended as an exchange format and not necessarily for querying. Screening of alterna-

tives led to the decision to implement the Clinical Quality Language (CQL) [21]. With CQL,

the backend can answer all queries generated by the Sample Locator. The implementation

started early in 2019 by defining biobank and biosample specific profiles based on the previ-

ously consolidated data set and publishing them on the open platform for HL71 profiles [22].

Successful query results delivered from the Sample Locator need to be communicated to

the respective biobanks for further refinement and finally to access to biosamples and data.

This is realized by an already developed IT-component, the BBMRI-ERIC Negotiator [23]. By

connecting this communication platform to the Sample Locator, any researcher, who has reg-

istered via the integrated authentication service BBMRI-ERIC AAI (Authentication and

Authorization Infrastructure), may initiate further communication with the respective bio-

banks [24].

To round up the data processes with and within the biobanks, the contact and consent

management with the sample donors needed to be considered as well. During a workshop

with patient representatives on this topic, it became evident that patients would not necessarily

contact the biobank but rather the clinical institution where the patient was treated and where

the biosample was taken. Based on this insight, a component for donor empowerment needs

to be embedded into the clinical environment, e.g., as part of an electronic health record

(EHR)-integrated patient portal instead of the biobank. For this reason, we recommend for

sustainable implementation to rather focus on industrial EHR-vendor-based developments.

However, a demonstrator that includes a patient portal with consent functionality and a con-

sent management database, an ID management tool, a project management tool, and an exem-

plary research database has already been piloted and can be built upon [25].

A detailed description of the IT infrastructure, its components and the underlying technol-

ogy is the subject of a separate manuscript, which is prepared parallel to this manuscript.

Sustainability

To ensure the sustainability of our results, all developments are made available as open source

under a general public license, so that interested parties can easily access and benefit from the

provided IT components. These are available in a GitHub repository set up for this purpose.

Moreover, an open-source community resulting from this project is supposed to keep the pub-

lished code maintained, modified and updated beyond the end of the project. This community

is open to everyone and should grow steadily in its further existence. A corresponding declara-

tion was formulated for this purpose [26].

Local implementation

After all legal requirements had been accomplished and the IT components were deployed, the

final step was to set up and operate the infrastructure on site in the respective biobanks by the

members of the local IT team. This step included on the one hand the integration of the IT

components into the biobank infrastructure and on the other hand the provision of the bio-

bank data via the connection to the required source systems. This was realized via an ETL

route specifically developed at each biobank which translates the source data into the specified

data format. This phase was particularly characterized by close cooperation between the two

IT teams. There was constant consultation and several iterations to adapt the implementation

and ETL to local conditions. Once this connection was successfully established using a

PLOS ONE Journey to establishing an IT-infrastructure for a biobank network

PLOS ONE | https://doi.org/10.1371/journal.pone.0257632 September 22, 2021 7 / 15

designated implementation guide [27], the respective biobank with its sample count was visible

in the Sample Locator for a researcher’s request for biospecimens (see also Fig 2).

Evaluation of components

The first usability analysis of the Sample Locator’s graphical user interface was based on a

mock-up and came to the conclusion that potential end users found the tool to be intuitive

and easy to handle [13]. A further usability analysis based on the productive system could con-

firm this impression [14]. Nonetheless, both analyses revealed useful potential for improve-

ment that could be incorporated into the subsequent development iterations.

By contrast, the evaluation of the patient portal demonstrator focused on the technical feasi-

bility of integrating it into the biobank infrastructure. By means of a checklist, the functionality

of the system was checked after the installation. The review of the completed checklists showed

that the implementation of such a portal is in principle technically possible. This finding is an

important input to increase the acceptance for the integration of such a portal into a hospital

infrastructure.

Discussion

The principal objective of our work within this project was to establish a functioning IT infra-

structure for networking biobanks, an endeavor that can be regarded as successfully accom-

plished. Nevertheless, several aspects arose in the course of the project that, in retrospect,

deserve special mention and attention (Table 1).

Challenges

Federated IT developments across locations turned out to be a challenge of its own [28]. The

spread of the teams across different sites, especially the development team, initially required a

certain orientation phase before beginning the actual work on the project. This proved to be

necessary since an efficient mode of operation had to be established first. After hiring of staff,

an agile development approach with daily scrum meetings and sprints was defined. Tools for

project management and a platform to document the progress of work supported this method.

Table 1. Short summary of challenges, lessons learned, and success stories we encountered during the GBA

project.

Challenges

• IT development across locations with heterogeneous background

• Coordination with other research infrastructures: communication and technical interoperability

• Clinic’s patient portal for consent management instead of a biobank specific donor portal

Lessons Learned

• IT development across locations with specific human resources

• Pursuit of open source development

• Comprehensive documentation & testing

• Alignment of the development with real world needs

• Continuous real-life monitoring (technically and manually) of the entire system

• Strong stakeholder involvement, e.g., user centered design approach and usability evaluations

Success Stories

• Early adaptation of HL7 FHIR to ensure interoperability of the biobank’s system

• Containerized deployment & detailed implementation guide

• Agile system development, e.g., enabled to offer specific access to COVID-19 samples online as early as April

2020

https://doi.org/10.1371/journal.pone.0257632.t001

PLOS ONE Journey to establishing an IT-infrastructure for a biobank network

PLOS ONE | https://doi.org/10.1371/journal.pone.0257632 September 22, 2021 8 / 15

In addition to this regular interaction, a chat tool has also been successfully used as a means of

asynchronous communication. When setting up the working environment, it is particularly

important to pay attention to the regulatory framework of the individual biobank/hospital

locations, as certain software, e.g., Zoom for video conferencing or Docker for container vir-

tualization, may not be permitted. Moreover, it should be ensured that all team members have

access to the source code and the corresponding development environment. This can be a bot-

tleneck especially if it is hosted at one of the sites and managed only by this site’s local admins.

Another aspect that can quickly be neglected in a distributed workspace is the communication

and transparency of decisions. In our specific case, this was aggravated by distribution of the

project responsibility among two of the central IT sites. Consequently, some changes in the

course of the project or with regard to the prioritization of work packages were passed on with

some delay.

A challenge that has arisen from external circumstances was and still is the coordination

with other research infrastructures, in particular the European BBMRI-ERIC and the German

MII. Due to partially overlapping goals within different settings, this reconcilement was essen-

tial to avoid the development of parallel structures. With regard to BBMRI-ERIC, significant

synergies could be created. The cooperation in terms of AAI [24], the Negotiator [23], and the

dataset based on MIABIS [18] led to a rather quick technical integration of the tools within

our IT infrastructure and resulted in international visibility and a high degree of interoperabil-

ity. However, one limitation of this adoption, which should be kept in mind, is the possible

limitation in functionality as in the case of the Negotiator as project management tool. At the

moment, it is limited to a chat-like communication between biobank and researchers in con-

trast to the targeted project proposal management tool.

On the other hand, such cooperation can also lead to the components being raised to

another level, as in the example of the donor portal, which is now being continued at hospital

level with a new funding call (Digitale FortschrittsHubs Gesundheit; digital progress hubs) of

the MII [29]. Another field concerns the coordination of the definition and integration of data

sets to ensure interoperability of the collected data. Towards this goal, the MII is especially

active in developing a standardized core data set for data integration centers since 2019 and

e.g., the definition of the biosample data module was therefore closely coordinated with the

respective MII initiative. Thus, we have been successful to interact with most relevant task

forces to contribute with our expertise, especially in the field of biobanking. However, it is still

crucial to make work and developments among the activities transparent and to communicate

with each other.

Lessons learned

Recruiting personnel for a project of this dimension is crucial to reach the full potential. Espe-

cially for those locations with a need for new recruitments there is an obvious delay, whereas

locations that can build on already available experts are in a better situation. Furthermore,

there was initially no coordination planed between the locations of the IT development team

regarding the competencies that would have been needed for this project. Particular attention

should be paid here to ensuring that certain areas are covered. Within this project, it con-

cerned the following fields: system administration, back-end and front-end development, user

interface design and operations. In addition, a person with experience in project management,

who has an overview of all work packages, is of utmost importance. Due to the partial lack of

coverage of the various expertise, additional support from some biobanks was required. Before

starting a project, we suggest considering to what extent the designated team members or tem-

porary interns can or could contribute to the tasks at hand.

PLOS ONE Journey to establishing an IT-infrastructure for a biobank network

PLOS ONE | https://doi.org/10.1371/journal.pone.0257632 September 22, 2021 9 / 15

A further difficulty was the partial fluctuation of employees almost from the beginning. In

particular, the small number of developers amongst the IT development team suffered by fric-

tion losses. The attempt to apply pair programming partially healed the situation. However,

some expertise was lost and the synchronization as a team was distracted. The changes in staff

also affected the rest of the team including the management level and thereby put a strain on

the project’s proceeding. However, the motivation within the team helped to overcome this

obstacle, even attracting IT developers from outside the original IT team.

In addition to human resources, the availability of infrastructure can also become a bottle-

neck. Here it should be ensured that a stable environment is provided. Of particular impor-

tance is a stable server infrastructure that is permanently available in terms of employee access

rights and performance. After initially using a server hosted by one of the IT competence cen-

ters for the project, we eventually switched to a cloud server. This enabled us to become inde-

pendent of location-immanent restrictions with regard to admin rights and ensures continued

maintenance. With this approach, though, sufficient funds should be earmarked for the main-

tenance of such an external service.

The postulate of sustainability, especially the publication of the software code as open-

source that was a rightful claim of the funding body and has been recognized as essential,

turned out to be rather time and resource consuming. This was mainly due to the fact that

work within this project was partially based on already existing software components. For

open source publication, there are different license types that can be used. Here, a decision was

required as to which of these types should be used for any further publication. The clarification

required legal advice and a consensus process between the project team and the legal represen-

tatives of the various universities involved. The effort and time required for such a process

were unfortunately underestimated at the start of the project. To this end, we strongly recom-

mend that all relevant legal topics should be as far as possible identified and consented already

by the start of a project. If existing code cannot be published as open source in the first months

one should consider alternatives–e.g., we have developed both the store and the GUI of the

Sample Locator from scratch and could thus easily publish them as open source.

In principle, it is favorable to use existing code in order to achieve higher productivity.

However, the development speed highly depends on the chosen code base, which should sat-

isfy common standards such as an up-to-date architecture, clean code, and reliable automated

tests. For example, adopting a hard to read software that is insufficiently covered by automated

tests extends the risk of errors—thereby imposing unnecessary workload on the team—and

slows down developers while writing and reviewing code. Moreover, a monolithic structure

that is used by more than one development team demands for proactive coordination. Hence,

analyzing the code base right at the beginning and being open to switch to another code base

or even start from scratch can drastically increase the overall performance of the development

team.

In IT-development, it is most crucial to stay connected with the requirements from actual

use cases and the stakeholders, especially the users of the proposed tools. Since our project was

not aiming merely at connecting biobanks across different locations, but also to enable specific

queries for biosamples to fuel scientific work, this needed to be outlined thoroughly. For this

purpose, we collected “real world” sample requests to test the system and realized that these

requests were by far more complex than anticipated. This was complicated by false positive

results that disclosed discrepancies as the donor’s age and the sample-derived diagnosis not

matching though both were entered for the search. This was detected by constant development

and technical as well as manual testing. This “real-life” monitoring definitely can be seen as

one valuable lesson learned.

PLOS ONE Journey to establishing an IT-infrastructure for a biobank network

PLOS ONE | https://doi.org/10.1371/journal.pone.0257632 September 22, 2021 10 / 15

Another aspect that this “real-life” monitoring revealed was that both frontend design and

user-friendliness had been underestimated within the project plan. The complexity of the

“simple” sample requests, gathered to test the system, was not easily covered by building a

generic user interface. Though the backend using CQL imposed no restrictions for the “real-

life” queries defined by researchers, the generic approach to configure the frontend using a

metadata repository and implementing simple logic on fields, values, and operators failed in

this regard. To make up for this omission, the team took part in a usability workshop and

applied the acquired knowledge in a subsequent evaluation of a sample locator prototype [13],

the results of which could be used in further development. Another lesson learned regarding

the graphical interface concerned the framework. The first attempt was built on the existing

code using Java Server Faces which was hard to maintain and almost impossible to enhance

with new features. We therefore decided to switch to a proper web framework (Angular)

because it is commonly used and suitable also for developers who are not specialized in web

development. In order not to neglect the cooperate design of the project, the GBA’s public rela-

tion coordinator supported the migration and development. Considering this, we recommend

to not underestimate the user’s experience in academic IT-developments since user interfaces

that are intuitive and well suited for the end user increase the acceptance of the tools and

thereby ensure the success of a project beyond its end [30].

Success stories

One major success of this project was the early adaptation of the HL71FHIR1 standard to

enhance interoperability. Originally, the HL7 standard was developed, among other things, for

the communication of patient data within a hospital and has become widely established. For

intersectoral communication, on the other hand, the relatively new FHIR standard was devel-

oped, so it was not yet widely used in academic research, at least in the biobanking community.

However, since this community operates at the interface between patient care and biomedical

research, we decided to adopt it for our purpose, namely saving and querying biobank data.

Accordingly, we have translated the metadata of the sample, donor, and biobank data set into

so-called FHIR profiles [22]. With those, the relevant data can now be easily standardized

within the established IT-infrastructure to make them easier to query. Although the decision

to follow this path was made relatively late in the project, it was consistently pursued by the

cooperation of the team by swiftly creating profiles of the biobank data set approved by the

community. The smooth implementation of this relatively new standard in biobanks was

favored by several factors: 1) no established format for cross-site communication in the bio-

bank community, so FHIR was able to fill this gap without displacing existing standards while

adding value to the given biobanking standards MIABIS and SPREC that were consulted and

respected during profiling, 2) the commitment of the project partners to introduce this largely

unfamiliar standard locally in the short term, 3) the general developments within the MII, in

which all German university hospitals also agreed to use the FHIR standard, and 4) the fact

that FHIR is structurally based on established HL7 standards, namely version 2 and CDA

(Clinical Document Architecture), and that the aim is to harmonize FHIR and CDA, making

mapping straightforward. As part of the decision to switch to FHIR, we also developed and

delivered a completely new local DWH, the Blaze Store [31], within a short period of time.

Although this meant additional effort, it also brought the advantage that the new store could

be tailored for the FHIR standard. This not only represents an additional benefit for GBA, but

also is currently being considered for use in other projects, including MII.

The activities of GBA also attracted the interest of other biobanks, which eventually applied

to join the alliance as partners [32]. Provided they met a predefined minimum requirement

PLOS ONE Journey to establishing an IT-infrastructure for a biobank network

PLOS ONE | https://doi.org/10.1371/journal.pone.0257632 September 22, 2021 11 / 15

[33], they were accepted as active members of the network. With regard to the delivery of the

IT tools, this posed a particular challenge since not all biobanks have their own IT staff. It was

therefore all the more important to offer an approach that makes it as straightforward as possi-

ble for biobanks to deploy the required tools. The decisive factor to achieve this was the deci-

sion to deliver the tools in a containerized version and to publish a detailed implementation

guide [27, 34]. Furthermore, the new members benefited from the experience of the trained IT

staff, who were able to pass on approaches to solving emerging problems expeditiously via the

established communication channels. Like this, 14 biobanks could already be connected to the

Sample Locator in a short time.

Effective dissemination of knowledge during the development of the IT infrastructure

allowed to quickly react to recent developments. For example, only just released for productive

use at the end of 2019, the IT infrastructure of German biobanks had to withstand a first baptism

of fire in 2020 with the outbreak of the COVID-19 pandemic and the soaring demand of

researchers for high-quality SARS-CoV-2 biosamples. Once the collection and storage of varie-

ties of these specific samples from biobanks had begun, the samples could be found immediately

online by searching for the International Classification of Diseases (ICD) 10 codes U07.1

(COVID-19) and U07.2 (suspected COVID-19) within the Sample Locator throughout Germany

(Fig 3). The German Biobank Node collected requests for SARS-CoV-2 specific samples (and

other samples) to track efficacy of the Sample Locator and Negotiator. From launch until today 7

SARS-Cov-2-specific sample requests for different sample types with or specifically without men-

tioned diagnosis have been received of which 4 requests could be mediated successfully.

Conclusion

The use of high quality biosamples and their associated clinical data from biobanks are essen-

tial to produce successful, reliable and reproducible biomedical research data. To meet this

Fig 3. Samples available in the Sample Locator as early as 28th of May 2020 from donors diagnosed with U07.1 or

U07.2 (COVID-19).

https://doi.org/10.1371/journal.pone.0257632.g003

PLOS ONE Journey to establishing an IT-infrastructure for a biobank network

PLOS ONE | https://doi.org/10.1371/journal.pone.0257632 September 22, 2021 12 / 15

demand, we have started to create an IT network of German biobanks that enables feasibility

queries across all connected sites. This was only possible by the efforts of a federated T devel-

oper team which was active on behalf of the German Biobank Node/Alliance. However, collab-

oration in such a distributed team requires the consideration of several prerequisites that need

to be taken into consideration before starting the project. In this paper, we want to share our

experience which might be of value for similar endeavors, and which should help to prevent

unnecessary mistakes (Table 2).

In fact, we are already able to benefit from our success and experience gained in this GBN/

GBA project for the biobank IT activities within the ABIDE_MI (Aligning Biobanking and

DIC Efficiently) project [35]. This is a successor project that has emerged from the previous

cooperation between MII and GBN and even includes more biobank sites.

Acknowledgments

We sincerely thank all active and former members participating in the IT development team

and the local IT team members of the German Biobank Alliance. Special thanks to Martin

Lablans whose relentless fight for headsets remains unforgotten.

The present work was performed in fulfillment of the requirements for obtaining the degree

“Dr. rer. biol. hum.” from the Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU) (CS).

The German Biobank IT development team over the term of the project: Martin Breu,

Noemi Deppenwiese, Christoph Dolch, Petra Duhm-Harbeck, Lars Ebert, Cäcilia Engels, Jori

Kern, Alexander Kiel, Christian Knell, Ann-Kristin Kock-Schoppenhauer, Björn Kroll, John

Linde, Christian Maier, Michael Neumann, Matthias Öfelein, Matthias Rambow, Susanne

Sahr, Christina Schüttler (lead author), Florian Stampe, Deniz Tas, Hannes Ulrich, as well as

Martin Lablans and Hans-Ulrich Prokoschas IT team leaders. Michal Hummel heads GBN

and the GBA project.

Author Contributions

Conceptualization: Hans-Ulrich Prokosch, Michael Hummel, Martin Lablans.

Project administration: Michael Hummel.

Software: Björn Kroll.

Supervision: Hans-Ulrich Prokosch, Michael Hummel.

Writing – original draft: Christina Schüttler, Björn Kroll, Cäcilia Engels.

Writing – review & editing: Christina Schüttler, Hans-Ulrich Prokosch, Michael Hummel,

Björn Kroll, Cäcilia Engels.

Table 2. Six key recommendations for similar projects.

1. Clarify legal conditions! Issues such as open source licensing should be settled with all affected parties before or

early in the project, in written form if necessary.

2. Recruit appropriate staff! Specify what expertise is needed and coordinate with project partners when staffing the

project.

3. Designate one permanent team leader! This person keeps track of project milestones and distributes precise tasks

according to their urgency to team members with the adequate skill.

4. Pay attention to areas of responsibilities! The team receives work assignments from the team leader, who in turn

communicates with the project management. Unclear hierarchies can be misleading.

5. Be flexible, but thoughtful! New conditions/requirements/technologies should be addressed, but not rashly.

6. Communicate! Involve your team in decisions and listen to its thoughts.

https://doi.org/10.1371/journal.pone.0257632.t002

PLOS ONE Journey to establishing an IT-infrastructure for a biobank network

PLOS ONE | https://doi.org/10.1371/journal.pone.0257632 September 22, 2021 13 / 15

References 1. Bernemann I, Kersting M, Prokein J, Hummel M, Klopp N, Illig T. Centralized biobanks: a basis for medi-

cal research. Bundesgesundheitsbl 2016, 59:336–343. https://doi.org/10.1007/s00103-015-2295-2

PMID: 26830106. [in German]

2. Pigeot I, Hummel M. Qualified biobanks are essential for successful biomedical research. Bundesge-

sundheitsbl 2016, 59:301–303. https://doi.org/10.1007/s00103-016-2307-x PMID: 26809820. [in

German]

3. Zatloukal K, Hainaut P. Human tissue biobanks as instruments for drug discovery and development:

impact on personalized medicine. Biomark Med 2010, 4:895–903. https://doi.org/10.2217/bmm.10.104

PMID: 21133710.

4. Chadwick D, Roehrl MHA. High-quality biobanking for personalized precision medicine: BioSpecimen

Sciences at the helm. Diagn Histopathol 2013, 19:447–456. https://doi.org/10.1016/j.mpdhp.2013.11.

009

5. Herpel E, Hummel M. Research biobanks–development and structure. Dtsch Med Wochenschr 2013,

138:1069–1072. https://doi.org/10.1055/s-0033-1343143 [in German] PMID: 23670264

6. Hummel M, Rufenach C. We’re here to accelerate Biomedical Research. 5. Nationales Biobanken

Symposium, Berlin 12/2016. ISBN 978-3-89838-721-7.

7. Prokosch HU. Pilotprojekt zur Biobanken-Vernetzung in GBN. 5. Nationales Biobanken Symposium. 5.

Nationales Biobanken Symposium, Berlin 12/2016. ISBN 978-3-89838-721-7.

8. Schmitt S, Becker KF, Meinung B, Herpel E, Slotta-Huspenina J. German Biobank Node´s concept for

quality management. Europe Biobank Week, Wien 9/2016.

9. Litton JE. Launch of an Infrastructure for Health Research: BBMRI-ERIC. Biopreserv Biobank 2018,

16:233–241. https://doi.org/10.1089/bio.2018.0027 PMID: 29781706.

10. Cohn M. Succeeding with agile. Software Development Using Scrum. Upper Saddle River, N.J: Addi-

son-Wesley, 2010. ISBN-10: 0321579364.

11. Lablans M, Kadioglu D, Muscholl M, Ückert F. Exploiting Distributed, Heterogeneous and Sensitive

Data Stocks while Maintaining the Owner’s Data Sovereignty. Methods Inf Med 2015, 54:346–352.

https://doi.org/10.3414/ME14-01-0137 PMID: 26196653.

12. Brooke J. SUS: a quick and dirty usability scale. In: Jordan PW, Thomas B, Weerdmeester BA, McClel-

land IL, editors. Usability Evaluation in Industry. London, England: Taylor and Francis; 1996: 189–194.

13. Schüttler C, Huth V, von Jagwitz-Biegnitz M, Lablans M, Prokosch HU, Griebel L. A Federated Online

Search Tool for Biospecimens (Sample Locator): Usability Study. J Med Internet Res 2020, 22:e17739.

https://doi.org/10.2196/17739 PMID: 32663150.

14. Schüttler C, Prokosch HU, Sedlmayr M, Sedlmayr S. Evaluation of the Feasibility Tools MIRACUM

i2b2, OHDSI’s ATLAS and GBA’s Sample Locator: A Comparative Usability Study. JMIR Med Inform

2021, 9:e25531. https://doi.org/10.2196/25531 PMID: 34287211

15. Semler SC, Wissing F, Heyder R. German Medical Informatics Initiative. Methods Inf Med 2018, 57:

e50–e56. https://doi.org/10.3414/ME18-03-0003 PMID: 30016818.

16. GBA Sample Locator. Available from: samplelocator.bbmri.de. [accessed 2021-05-14]

17. Schüttler C, Buschhüter N, Döllinger C, Ebert L, Hummel M, Linde J, et al. Requirements for a cross-

location biobank IT infrastructure: Survey of stakeholder input on the establishment of a biobank net-

work of the German Biobank Alliance (GBA). Pathologe 2018, 39:289–296. https://doi.org/10.1007/

s00292-018-0435-9 PMID: 29691676. [in German]

18. Eklund N, Andrianarisoa NH, van Enckevort E, Anton G, Debucquoy A, Müller H, et al. Extending the

Minimum Information About BIobanking data Sharing -terminology to describe samples, sample donors

and events. Biopreserv Biobank 2020, 18: 155–164. https://doi.org/10.1089/bio.2019.0129 PMID:

32302498.

19. ISBER Biospecimen Science Working Group. Standard PREanalytical Code (SPREC). Available from:

https://www.isber.org/page/SPREC [accessed 2021-05-14]

20. Welcome to FHIR. HL7 International. Available from: https://www.hl7.org/fhir/ [accessed 2021-05-14]

21. HL7 International. CQL. Clinical Quality Language Release 1. Available from: https://cql.hl7.org/

[accessed 2021-05-14]

22. BBMRI.de. Profiles for the BBMRI.de / GBA biobanking project. Available from: https://simplifier.net/

bbmri.de/~resources?category=Profile [accessed 2021-05-14]

23. Reihs R, Proynova R, Maqsood S, Ataian M, Lablans M, Quinlan PR, et al. BBMRI-ERIC Negotiator:

Implementing Efficient Access to Biobanks. Biopreserv Biobank 2021 (forthcoming). https://doi.org/10.

1089/bio.2020.0144 PMID: 34182766

PLOS ONE Journey to establishing an IT-infrastructure for a biobank network

PLOS ONE | https://doi.org/10.1371/journal.pone.0257632 September 22, 2021 14 / 15

24. CESNET, CERIT-SC, Masaryk University. PERUN-AAI. Available from: https://perun-aai.org/

25. Öfelein M, Reichold M, Maier C, Prokosch HU. Designing a Framework of Components to Support

Patient Engagement in Research. Stud Health Technol Inform 2019, 267:20–27. https://doi.org/10.

3233/SHTI190800 PMID: 31483250.

26. Samply Community. Available from: https://github.com/samply [accessed 2021-05-14]

27. BBMRI.de. GBA Implementation Guide. Available from: https://samply.github.io/bbmri-fhir-ig/

[accessed 2021-05-14]

28. Sahr S, Kern J, Tas D, Döllinger C, Duhm-Harbeck P, Ebert L, et al. Biobanken-Vernetzung mittels

Scrum in einem Standort verstreuten IT-Team. 6. Nationales Biobanken-Symposium 2017; 111–113.

ISBN: 978-3-89838-730-9.

29. Medical Informatics Initiative. „Digitale FortschrittsHubs Gesundheit“. Available from: https://www.

gesundheitsforschung-bmbf.de/de/Digitale-FortschrittsHubs-Gesundheit.php [accessed 2021-05-26]

30. ISO 9241–210:2019 Ergonomics of human-system interaction—Part 210: Human-centred design for

interactive systems. Geneva, Switzerland: International Organization for Standardization; 2019. Avail-

able from: https://www.iso.org/standard/77520.html/ [accessed 2021-05-14]

31. Samply Community. Blaze. Available from: https://github.com/samply/blaze [accessed 2021-05-14]

32. BBMRI.de. Become a partner. Available from: https://www.bbmri.de/about-gbn/become-a-partner/?L=

1 [accessed 2021-05-14]

33. BBMRI.de. Requirements catalogue for biobank candidates. Available from: https://www.bbmri.de/

fileadmin/user_upload/PDFs/Fragenkatalog_Partner-Biobank_GBA_2020.pdf [accessed 2021-05-14]

34. Samply Community. Bridgehead Deployment. Available from: https://github.com/samply/bridgehead-

deployment/ [accessed 2021-05-14]

35. Medical Informatics Initiative. ABIDE_MI. Available from: https://www.medizininformatik-initiative.de/

en/node/609 [accessed 2021-05-14]

PLOS ONE Journey to establishing an IT-infrastructure for a biobank network

PLOS ONE | https://doi.org/10.1371/journal.pone.0257632 September 22, 2021 15 / 15