DVCS in Europe PMC OA articles

dvcs_url.csv

rebiDVCS.md

Aim

The aim of this post is to track the use of Open-Source software hosting facilities for disclosing DVCS repositories in life science research papers.

Data and Methods

Our sample consists of the following four hosting services provided by Wikipedia

my.urls <- read.csv("dvcs_url.csv", header = TRUE, sep = ",")
print(my.urls)

##          dvcs        base.url
## 1      GitHub      github.com
## 2   BitBucket   bitbucket.org
## 3  GoogleCode code.google.com
## 4   LaunchPad   launchpad.net
## 5 SourceForge sourceforge.net

We used Europe PMC as literature corpus. We searched for the url patterns within the publication subset for which Europe PMC holds the full text. For this, we used the rebi package, provided by rOpenSci.

require(rebi)
require(plyr)
my.urls$base.url <- as.character(my.urls$base.url)
my.data <- lapply(my.urls$base.url, search_publications, dataset = c("fulltext"))
names(my.data) <- my.urls$base.url
my.data <- ldply(my.data, rbind)

Results

We have found

length(unique(my.data$id))

[1] 3190

publications referencing at least one Open-Source software hosting service.

Table 1 ranks the host services

id	PMC Publications found
sourceforge.net	1495
code.google.com	914
github.com	832
bitbucket.org	88
launchpad.net	16

Figure 1 plots the yearly distribution of DVCS hosting sservcies over PubMed Central publications. Please note that data were gathered on

[1] "2014-05-23 16:12:24 CEST"

require(ggplot2)

my.data <- my.data[my.data$pubYear > 2008 & my.data$pubYear < 2014, ]
my.data$.id <- factor(my.data$.id, levels = c(rownames(data.frame(rev(sort(table(my.data$.id)))))))

my.df <- data.frame(as.matrix(table(unlist(my.data$pubYear), my.data$.id)))

ggplot(my.df, aes(Var1, Freq, group = Var2)) + geom_line(aes(colour = Var2, 
    show_guide = FALSE)) + geom_point() + theme_bw() + scale_colour_brewer("DVCS Host", 
    palette = 2, type = "qual") + xlab("Year") + ylab("PMC article disclosure") + 
    opts(legend.key = theme_rect(fill = "white", colour = "white"))

Discussion

We have found that GitHub is gaining in importance for data and code disclosure in the life sciences compared to other DVCS hosting services.

rebiDVCS.Rmd

## Aim

The aim of this post is to track the use of Open-Source software hosting facilities for disclosing DVCS repositories in life science research papers.

## Data and Methods

Our sample consists of the following four hosting services provided by [Wikipedia](http://en.wikipedia.org/wiki/Comparison_of_open-source_software_hosting_facilities)

```{r, warning = FALSE, message = FALSE}
my.urls <- read.csv("dvcs_url.csv", header = TRUE, sep = ",")
print(my.urls)
```

We used Europe PMC as literature corpus. We searched for the url patterns within the publication subset for which Europe PMC holds the full text. For this, we used the [rebi](http://ropensci.github.io/rebi/) package, provided by [rOpenSci](http://ropensci.org).

```{r, warning = FALSE, message = FALSE}
require(rebi)
require(plyr)
my.urls$base.url <- as.character(my.urls$base.url)
my.data <- lapply(my.urls$base.url, search_publications, dataset = c("fulltext"))
names(my.data) <- my.urls$base.url
my.data <- ldply(my.data, rbind)      
```

## Results 

We have found 
```{r , results='asis', warning = FALSE, message = FALSE}
length(unique(my.data$id))
```
publications referencing at least one Open-Source software hosting service. 

Table 1 ranks the host services

```{r , results='asis', echo=FALSE, warning = FALSE, message = FALSE}
my.table <- data.frame(sort(table(my.data$.id), decreasing =T))
colnames(my.table) <- "PMC Publications found"
kable(my.table, format = "markdown")
```

Figure 1 plots the yearly distribution of DVCS hosting sservcies over PubMed Central publications. Please note that data were gathered on

```{r , results='asis', echo=FALSE}
print(Sys.time())
```


```{r simpleplot, echo=TRUE, fig.height=8/2.54, fig.width=18/2.54, warning = FALSE, message = FALSE}
require(ggplot2)

my.data <-  my.data[my.data$pubYear > 2008 & my.data$pubYear < 2014,]
my.data$.id <- factor (my.data$.id, levels = c(rownames(data.frame(rev(sort(table(my.data$.id)))))))

my.df <- data.frame(as.matrix(table(unlist(my.data$pubYear), my.data$.id)))

ggplot(my.df, aes(Var1, Freq, group = Var2)) + 
  geom_line(aes(colour = Var2, show_guide=FALSE)) + geom_point() + theme_bw() + 
  scale_colour_brewer("DVCS Host",palette=2, type="qual") + 
  xlab("Year") + ylab("PMC article disclosure") +
  opts(legend.key=theme_rect(fill="white",colour="white"))

```
## Discussion

We have found that [GitHub](http://github.com) is gaining in importance for data and code disclosure in the life sciences compared to other DVCS hosting services.