Creating spectral database for biomedical applications
Contents
1. Biomedical diagnostics based on Vibrational
spectroscopy
6. Authorship & data
ownership
Despite the
advancement in molecular biology, biomedical diagnostics still face many
challenges especially in resource-limited settings, which highlights the need
for an efficient and cost effective alternatives [1]. The potential of vibrational spectroscopy in
providing such alternative has been unearthed by the development of diverse
multivariate data-analysis techniques, which allowed efficient chemometric
analyses. These techniques allowed better utilization of the information-rich
spectra produced by diverse spectroscopic techniques, including vibrational
spectroscopy. The spectra produced by vibrational spectroscopy reflects the
chemical composition of the analyzed samples, which could be used for molecular
fingerprinting due to the unique absorption patterns for each constituent,
which enable direct identification at a molecular level. The spectral
fingerprint could be analyzed and assigned by chemometric analysis using the
peaks position and intensities as multivariate data [2].
The diverse
vibrational spectroscopic techniques share common advantages over the current
traditional diagnostic techniques. These advantages include the extreme low
running cost, the simple procedures and limited sample preparation, the
environment-friendly procedure (green chemistry), the time efficiency and the
potential portability of some of these spectroscopic techniques (such as Infrared
spectroscopy and Raman spectroscopy)[1-3].
Thus,
vibrational spectroscopy has been investigated for diverse biomedical
applications including diagnosis of cancers, metabolic diseases (such as
diabetes), inflammatory diseases, and infectious diseases [1, 2, 4]. Moreover, multivariate analysis,
also, allowed quantitative analysis of drugs by vibrational spectroscopy in
plasma without extraction [5], which highlights its potential for cost-effective
therapeutic drug monitoring.
To assess and
translate the potential of biospectroscopy in
clinical diagnostics and therapeutic Drug Monitoring.
SA1: to create a comprehensive spectral database for biomedical
applications using simulated clinical samples & de-identified leftovers
with real-world data.
SA2: to assess the sensitivity/specificity of IR, NIR, Raman UV spectroscopy
in the diagnosis of multiple disease & in TDM using this database in
controlled clinical trials
.
4.1 SA1: Creating
a comprehensive spectral database for biomedical applications
4.1.1 Spectral
data creation
This part of
the project has 3 arms Clinical simulation, Clinical Studies, Real-world data.
4.1.1.1 Clinical
simulation
This include spiking
pooled human sera with known biomarkers, etiological agents of infectious
diseases, or drugs (both qualitative and quantitative). Then, the spectrum (IR,
NIR, Raman, or UV) of these spiked sera and free serum (as control in the same
conditions, where the serum is divided before into those to be spiked and those
to act as control) is to be collected directly after efficient mixing.
The data
from this arm, although sufficient to publication, is just to provide to
provide complimentary information. This is because the real-world samples are
much complicated and in fact multiple factors can interfere with spectral data,
not just a single factor, which all should be addressed collectively.
4.1.1.2 Real-world data
De-identified clinical biopsies (mainly plasma, and blood) would be
provided from the clinical partners. They will be labeled with relevant and
rich clinical information (which exclude any information that can lead to the
identity of the patient) and included in a database with unique ID.
Then these
biopsies will be scanned (at least three times) by the physical partners and
the spectral data & the method of scanning (including the device, the type
of material used, and any other conditions) will be linked to the clinical data
in the database.
In this case
any sample with the disease/drug/or any other factors studied will be selected
as a test and the others remaining will be considered control.
This is the
most significant part of the project as it provides the most data for the
library and will aid in assessing the real value of biospectroscopy
in clinical practice.
4.1.1.3 Clinical/animal
studies
This include
obtaining clinical biopsies from patients enrolled in clinical studies where
patients with the disease are being the test and others are the control. TDM
can be performed on animals then assessed in human in typical reported trials
(concentration of the drugs assessed by a reference method such as HPLC and
used to create a model of the new method using chemometric regression
analysis). They will be labeled and scanned like the samples from real-world
data arm, and the data will be included in the database.
Although this
arm is the most common way approached by the scientists working in the field of
clinical spectroscopy and it is likely to provide the most accurate result,
this arm will requires more time for regulatory
approvals, patient enrollment (especially in rare diseases) and would provide
much less data than real-world data, that could lead to less significance of
the result. Thus, this arm will only be approached if feasible and preferably
in assessing the database not in creating it.
4.1.2 Data management and analysis
The data
science partner will be responsible for data management and analysis. This
include both creating the database (clinical, spectral, method data), and
preprocessing and chemometric analysis of the spectral data using R and the
specialized software Unscrambler. Of course, data will be available for all
partners to perform the analysis they want.
It should be
noted that there are multiple chemometric analysis that can be approached to
reach a model effective for clinical application of vibrational spectroscopy.
This also include diverse preprocessing of the spectra. Moreover, such project
is a big data project, that needs many data to reach significance. For that reason, it is desirable to allow the
data form this project to be available open access to allow data sharing and
compiling of data from diverse partners for a bigger data and better
significance, putting in consideration the authorship rights of all those
participated in the project.
4.2 SA2:
to assess the sensitivity/specificity of biospectroscopy
in the diagnosis of multiple disease & in TDM using this library
Upon
reaching a model with significant accuracy and sensitivity for a specific
disease diagnostic or TDM, this model will be challenged by the new samples to
identify such disease or concentration of the medication.
The model
showing enough accuracy and sensitivity will also be assessed as a novel
diagnostic technique in clinical studies.
4.3 Main
Project Reference
All partners
of this project have the right to use any method to perform their part of the
project, but they should mention in detail the method they used.
The main
reference to this project is baker et al 2014 nature protocol “Using Fourier
transform IR spectroscopy to analyze biological materials” attached as appendix
II to this proposal
5.1
Microbial strains
For clinical simulation studies.
5.2 Drugs
For clinical simulation studies.
5.3 KBR
For FTIR scanning
5.4 Equipment and Software
· Spectrophotometers (IR, UV,
Raman, etc)
· Lyophilizer
· Software for multivariate
analysis such as “Unscrambler” & R.
Every sample
analyzed will be assigned with at least three authors (clinical, physical,
& Data), any other authors will also be added for any contribution with
clear description of their contribution.
These
authors will be added in the database and will be available for all authors to
view at any time.
If any
spectral data used in any publication by any of the partners in this project,
they will be notified to be added as authors in the publication, with
description of their contribution (for responsibilities issues). First author
will vary and will always be the conceiver of the idea of that specific
publication or any other if she/he sees deserve to be the first author.
If data
published as open data, the data will be cited by others for their publication
and that will be mainly data users other than the partners.
If the data
created have any monetary value, the value of the data will be shared equally
to those who contributed to it. Each data record value will be divided into 3
equal shares: physical, clinical and data. The authors should discuss with
their respective institute their monetary ownership of their shares of the
data.
When
applying to any funding, all partners should be notified, and the PI should be
nominated among those of acceptable credentials especially those applying for
the funding.
Areas of
funding include expanding the scale of the project, purchasing new devices,
optimizing the infrastructure especially those needed for data sharing, and any
other aspects any partner consider appropriate.
This project
is interdisciplinary in nature including partners from the field of biophysics,
clinical field, & data science; the current partners are listed in Appendix
I
Due to the
nature of the project, addition of new partners is encouraged. However,
partners should be notified upon the addition of new partners; the reasons
should be discussed, and they include new devices, better availability, more
clinical data, etc. To further promote the activities of this project, upon the
1st 10 partners will discuss the establishment an association to
promote the activities of the partners and receive funds for these purposes,
which is suggested to be called “The Egyptian Network for Clinical
Spectroscopy”.
· Needed devices:
FTIR spectrometer (ATR would be preferred, others are of course
acceptable), or/& Raman spectrometer and to less extent UV
·
Nature of samples: blood and
plasma samples from cancer patients provided as liquid if ATR or Raman are used
and lyophilized in case of using kbr
o
For safety purposes HCV and HIV
samples will be excluded
·
Desired output from physical
partner: raw data (excel or text) so we can perform chemometric analysis on
them; large number of samples are needed for such analysis. We will need
information about the processing necessary to be done (such as
background correction)
·
Project nature: A long-term
project in the field of clinical spectroscopy (estimated 50-100 sample/month
each sample scanned 3 times)
·
The goal: providing a more cost
effective, fast and convenient diagnostics for multiple diseases including
diverse cancers
·
The project is intended to be long
term with multiple publications based on the results
o
Any participants will be included
as an author, participation is based on the data included in the study.
|
Version: |
2 |
|
Status: |
Draft |
|
Date of Issue: |
16/6/2019 |
|
Author: |
O.A Elkadi |
|
Summary of Changes: |
Version |
Date |
Description |
|
No changes |
1 |
26/3/2019 |
1st draft with expected amendments/additions |
|
1.Data Ownership detailed 2.Data Science partner added 3.Project outline amended 4.Project Flow chart added in the project
summary |
2 |
16/06/19 |
|
I. Current
Project Partners
II. Main reference to the methods in this project: Using Fourier transform IR spectroscopy to analyze biological materials