Content uploaded by Billy Bostickson
Author content
All content in this area was uploaded by Billy Bostickson on Oct 25, 2021
Content may be subject to copyright.
The second peak papers
Research by Anon1, Yuri Deigin, Billy Bostickson, and EricRQ
Curious findings in the metrics of published scientific papers
May, 2021
After publication of a scientific peer-reviewed paper it is to be expected that there is activity
reading abstracts, reading full text versions, and downloading pdfs of the paper for
reference.
During June 2020 we noticed that a few papers related to coronavirus research are not
following this expected development, but in addition to the first peak around the time of
publication has a second peak 2018/2019. At first the papers noticed peaked late in 2019.
After further work in 2021 we found that the second peak paper period runs from September
2018 to December 2019 – around 15 months in total. We have collected 22 papers so far
that we qualify as second peak papers.
The criteria are as follows:
● No or virtually no reads (full text/pdf) since publication until September 2018
● Minimum 150 reads during any month
For all the papers included the timeline since publication can be shown as the below:
Figure 1: All views since publication
It is the read activity just prior to January 2020 that is of interest. When only showing the
time period from 2018 to 2019 (excluding everything before 2018 and after 2019) the activity
can be shows as the below:
Figure 2: All views in 2018 and 2019, timeline/aggregated
As can be seen the characteristics (or “fingerprint”) of a second peak paper is generally:
● High and fluctuation abstract read coming down to virtually no reads late in 2018
● Thereafter either of three peaks; late 2018/early 2019, mid 2019, or late 2019
● Virtually no pdf downloads, nearly all full text views
● Most are published on jvi.asm.org
● Reads have a big uptake from virtually no reads in the preceding years to high
numbers, sometimes reaching 8-900 reads per month during the second peak
● Abstract reads flatten at near zero levels until May 2019 thereafter the reads peak
before falling again to very low levels late 2019
● The reads peaks in September or October 2019 before rapidly declining
When showing synchronicity of the paper reads, the analysis present a clear wave pattern:
Figure 3: All views 2018 and 2019, timelined/per paper
The views presented makes a few main points clear:
● There was an unexplained increase in reads from late 2018 to end of 2019
● The reads consists of six observations:
○ Virtually no reads for these papers prior to Quarter 3, 2018
○ Fluctuating high abstracts falling to near zero in December 2018
○ At the same time the full text/pdf reads paradoxically increases
○ Abstract reads are after several year long period of hugh and fluctuating read
down to virtually zero for a 4 month period in 2019, before peaking and then
declining
○ The reads of full text/pdfs are distributed in three distinct plateaus:
■ Quarter 3 to Quarter 1, 2019 - Plateau 1
■ Quarter 2, 2019 - Plateau 2
■ Quarter 3, 2019 - Plateau 3
○ The wave pattern is disturbed by two papers - one being the dominant read
until July/August 2019, and then falling to near zero in September/October
2019 - and the other suddenly dominating the reads from the same time
(July/August 2019) until September/October 2019 where this also falls to near
zero reads
● All reads rapidly decline from September/October, or Quarter 4, 2019 onwards
Interpretations of this must acknowledge the following:
● The data does not show where the reads took place
● The data does not show who read the papers
There are indications, however, in the data that gives some ideas:
● The consistent wave-like pattern indicates that it is the same group of readers
● The fall in abstracts and the increased read in plateaus indicate a coordinated effort
● The sudden change abstracts peaking at the same time as one paper is no longer
downloaded, while a new suddenly is, combined with a specific uptake in one
abstract; and a somewhat more blurry Quarter 4, 2019 indicates an event followed by
a response
Note: The data does not show what kind of event, but indicates that there was an anomaly in
June 2019 (+/- 1 month). But also note that the massive reads following June 2019 does
indicate a follow-up to this event.
The details:
● The dominant paper that suddenly dropped in July/August 2019 is
Receptor Usage and
Cell Entry of Porcine Epidemic Diarrhea Coronavirus, https://jvi.asm.org/content/89/11/6121
● The paper that the abstracts peaked for is
Discovery of Novel Bat Coronaviruses in South
China That Use the Same Receptor as Middle East Respiratory Syndrome Coronavirus,
https://jvi.asm.org/content/92/13/e00116-18
● The paper that became the dominant paper since July/August 2019 is Severe Acute
Respiratory Syndrome (SARS) Coronavirus ORF8 Protein Is Acquired from SARS-
Related Coronavirus from Greater Horseshoe Bats through Recombination,
https://jvi.asm.org/content/89/20/10532
The other peaking papers are listed below (papers noted in red has been dropped due to
few reads):
# Title Jorunal Year Reference
1
Enhanced Virulence Mediated by the Murine
Coronavirus, Mouse Hepatitis Virus Strain JHM, Is
Associated with a Glycine at Residue 310 of the Spike
Glycoprotein JVI 2003
https://jvi.asm.o
rg/content/77/1
9/10260/article-
info
2
A Single Amino Acid at the Hemagglutinin Cleavage
Site Contributes to the Pathogenicity but Not the
Transmission of Egyptian Highly Pathogenic H5N1
Influenza Virus in Chickens JVI 2013
https://jvi.asm.o
rg/content/87/8/
4786/article-
info
3
Recombinant Receptor-Binding Domains of Multiple
Middle East Respiratory Syndrome Coronaviruses
(MERS-CoVs) Induce Cross-Neutralizing Antibodies
against Divergent Human and Camel MERS-
CoVs and
Antibody Escape Mutants JVI 2016
https://jvi.asm.o
rg/content/91/1/
e01651-
16/article-info
4
A Single Point Mutation Creating a Furin Cleavage
Site in the Spike Protein Renders Porcine Epidemic
Diarrhea Coronavirus Trypsin Independent for Cell
Entry and Fusion JVI 2015
https://jvi.asm.o
rg/content/89/1
5/8077/article-
info
5
Isolation and Characterization of a Novel Bat
Coronavirus Closely Related to the Direct Progenitor
of Severe Acute Respiratory Syndrome Coronavirus JVI 2016
https://jvi.asm.o
rg/content/90/6/
3253/article-
info
6
Discovery of Novel Bat Coronaviruses in South China
That Use the Same Receptor as Middle East
Respiratory Syndrome Coronavirus JVI 2018
https://jvi.asm.o
rg/content/92/1
3/e00116-
18/article-info
7
Cleavage of Group 1 Coronavirus Spike Proteins: How
Furin Cleavage Is Traded Off against Heparan Sulfate
Binding upon Cell Culture Adaptation JVI 2008
https://jvi.asm.o
rg/content/82/1
2/6078/article-
info
8
Preventing Cleavage of the Respiratory Syncytial Virus
Attachment Protein in Vero Cells Rescues the
Infectivity of Progeny Virus for Primary Human Airway
Cultures JVI 2016
https://jvi.asm.o
rg/content/90/3/
1311/article-
info
9
A Polymorphism within the Internal Fusion Loop of the
Ebola Virus Glycoprotein Modulates Host Cell Entry JVI 2017
https://jvi.asm.o
rg/content/91/9/
e00177-
17/article-info
10
Severe Acute Respiratory Syndrome Coronavirus
Protein 6 Accelerates Murine Coronavirus Infections JVI 2007
https://jvi.asm.o
rg/content/81/3/
1220/article-
info
11
Systematic Assembly of a Full-Length Infectious cDNA
of Mouse Hepatitis Virus Strain A59 JVI 2002
https://jvi.asm.o
rg/content/76/2
1/11065/article-
info
12
Receptor Usage and Cell Entry of Porcine Epidemic
Diarrhea Coronavirus JVI 2015
https://jvi.asm.o
rg/content/89/1
1/6121/article-
info
13
Viral Expression of CCL2 Is Sufficient To Induce
Demyelination in RAG1−/− Mice Infected with a
Neurotropic Coronavirus JVI 2005
https://jvi.asm.o
rg/content/79/1
1/7113/article-
info
14
Receptor Variation and Susceptibility to Middle East
Respiratory Syndrome Coronavirus Infection JVI 2014
https://jvi.asm.o
rg/content/88/9/
4953/article-
info
15
Receptor Recognition Mechanisms of Coronaviruses:
a Decade of Structural Studies JVI 2015
https://jvi.asm.o
rg/content/89/4/
1954/article-
info
16
Severe Acute Respiratory Syndrome Coronavirus
Infection Causes Neuronal Death in the Absence of
Encephalitis in Mice Transgenic for Human ACE2 JVI 2008
https://jvi.asm.o
rg/content/82/1
5/7264/article-
info
17
Switching Species Tropism: an Effective Way To
Manipulate the Feline Coronavirus Genome JVI 2003
https://jvi.asm.o
rg/content/77/8/
4528/article-
info
18
Severe Acute Respiratory Syndrome (SARS)
Coronavirus ORF8 Protein Is Acquired from SARS-
Related Coronavirus from Greater Horseshoe Bats
through Recombination JVI 2015
https://jvi.asm.o
rg/content/89/2
0/10532/article-
info
19
The Coronavirus Spike Protein Is a Class I Virus
Fusion Protein: Structural and Functional
Characterization of the Fusion Core Complex JVI 2003
https://jvi.asm.o
rg/content/77/1
6/8801/article-
info
20
Structural Characterization of Human Coronavirus
NL63 N Protein JVI 2017
https://jvi.asm.o
rg/content/91/1
1/e02503-
16/article-info
21
Receptor Variation and Susceptibility to Middle East
Respiratory Syndrome Coronavirus Infection JVI 2014
https://jvi.asm.o
rg/content/88/9/
4953/article-
info
22
Middle East Respiratory Syndrome Coronavirus
Nonstructural Protein 16 Is Necessary for Interferon
Resistance and Viral Pathogenesis mSphere 2017
https://msphere
.asm.org/conte
nt/2/6/e00346-
17/article-info
23
A Mouse Model for Betacoronavirus Subgroup 2c
Using a Bat Coronavirus Strain HKU5 Variant mBio 2014
https://mbio.as
m.org/content/5
/2/e00047-
14/article-info
24
Human Coronavirus EMC Does Not Require the
SARS-Coronavirus Receptor and Maintains Broad
Replicative Capability in Mammalian Cell Lines mBio 2012
https://mbio.as
m.org/content/3
/6/e00515-
12/article-info
25
Tight Junction Protein Occludin Is a Porcine Epidemic
Diarrhea Virus Entry Factor JVI 2017
https://jvi.asm.o
rg/content/91/1
0/e00202-
17/article-info
Table 1: Overview of papers included
Note: At the time of writing (June 6, 2021) the jvi.asm.org website has changed and the
details can no longer be accessed.
The second peak papers - some questions.
June 9, 2021
Dear reviewer,
Thanks a lot for taking the time to review the presentation and for commenting on the idea.
It’s much appreciated. The backdrop to this is in 2020 when some readers of academic
papers noticed some curious peaks in downloads/reads of papers as shown in the graphs on
the publisher websites. As a few of these readers eventually connected we aggregated the
papers we had discovered and took the time to aggregate the findings. The main issue
raised is in our view valid and also incredibly difficult to get past.
We will set out an explanation here.
As illustrated below the uptake in reads at the end of 2019 is almost invisible.
Figure 1: View of all papers included since publication until May, 2021
When removing all reads after January, 2020 the table takes on a different character, and
the uptake that can be noticed just prior to 2020 becomes more curious..
Figure 2: All reads since publication until December 31, 2019
In order to make this clearer we additionally increased the granularity to only look at 24
month prior to January 1, 2020.
Figure 3: All reads between January 1, 2017 and December 31, 2019
As we were looking for a second peak since publication late in 2019, this “ridge” coming up
and then declining prior to the end of 2019, was what we were trying to figure out what was.
As work progressed we found that the peaks started earlier, around 3rd quarter 2018, or
around 15 months prior to the end of 2019.
When splitting the time period in two 18 months periods - a total of 36 months prior to the
end of December 2019, it looks like this:
Figure 4: All reads from January 2017 to June 2018
Figure 5: All reads from June 2018 to December 2019
This curious development is shown as the difference between Figure 4 and Figure 5. What is
it that makes papers that had a stable read base since 2013, continuing on Figure 4, to
suddenly peak late in 2019 as shown in Figure 5?
These are papers we have been looking for. And we have found around 21, and will soon
add about 10 more.
As you point out there are issues here with bias, and “baselining” of the findings.
As it is we thought the papers themselves would be the only possible baseline (as shown in
Figure 4).
The development seen in Figure 5 is not to be expected.
As shown in Figure 2, the reads climb up a wall, starting late in 2018 - peaking in September
and October 2019, until the reads rapidly decline the last few months of 2019.
The point is, however, valid. We took the following steps to reduce the effect of bias, and to
qualify the findings:
● We removed all papers not published prior to 2018 to set a baseline
● All papers included needed at least 150 reads in a month
●
As a control we looked at papers with general validity for coronavirus research, such as
“Receptor Recognition Mechanisms of Coronaviruses: a Decade of Structural Studies”, by
Fang Li, published in November 2014, JVI.
It is indeed curious for a foundational paper like this, with now nearly 54.000 reads since
publication, to suddenly peak late in 2019, ref. the figure below.
Figure 6: Receptor Recognition Mechanisms of Coronaviruses: a Decade of Structural
Studies”, by Fang Li, published in November 2014, JVI. All reads in 2018 and 2019.
As may be observed the paper follows the expected development and stays around its
average read numbers of less than 100 reads p/m, until it suddenly jumps in March 2019
and peaks a few months later, in August 2019, with more than 800 reads p/m. Also note the
rapid decline following September 2019, which holds true for nearly all of the second peak
papers.
We are currently working to timeline the reads as shown below. As is clear, the downloads
follow a pattern that is difficult to ignore.
Figure 7: Sparkline, all papers, all reads until December 31, 2019
To conclude, we do not know who read the papers, or where the readers were. We have not
been able to figure out why so many of the papers were published in the JVI (not all but most
are).
We do, however, see a clear pattern that indicates that it is the same group of readers we
are seeing the echoes of.
What is it that connects these papers?