Merge pull request #5 from UW-GAC/feature/add-dates

Add request and renewal dates to the output

.gitignore

@@ -161,4 +161,5 @@ cython_debug/
 
 ###########
 # Custom additions
-dars.tsv
+dars.tsv
+*.html

README.md

@@ -46,6 +46,6 @@ python parse_dars.py <path_to_application_pdf> <path_to_output_tsv>
 ## Updating the docker images
 
 ```
-docker build -t uwgac/parse-dbgap-application:<tag> .
+docker build --no-cache -t uwgac/parse-dbgap-application:<tag> .
 docker push uwgac/parse-dbgap-application:<tag>
 ```

dar_report.Rmd

@@ -46,3 +46,19 @@ tmp <- dars %>%
 nrow(tmp)
 tmp %>% kable()
 ```
+
+# Dates
+
+## Request date
+
+```{r}
+ggplot(dars, aes(request_date)) +
+    geom_histogram()
+```
+
+## Renewal date
+
+```{r}
+ggplot(dars, aes(last_renewal_date)) +
+    geom_histogram()
+```

parse_dars.py

@@ -7,6 +7,7 @@
 
 
 def parse_phs_blocks(blocks):
+    #import ipdb; ipdb.set_trace()
     phs = blocks[0]
     request_date_idx = [i for i, item in enumerate(blocks) if re.search("Request Date", item)]
     dar_list = []
@@ -23,6 +24,10 @@ def parse_phs_blocks(blocks):
             this_dar["DAR"] = m.string.split(" : ")[1]
         else:
             this_dar["DAR"] = None
+        # Get the request and renewal dates.
+        date_blocks = blocks[idx].replace(":\n", ": ").split("\n")
+        tmp = {k.strip(): v.strip() for k, v in (xx.split(":") for xx in date_blocks)}
+        this_dar.update(tmp)
         # Now find the consent group.
         j = idx
         try:
@@ -88,7 +93,9 @@ def parse_phs_blocks(blocks):
     # Convert to pandas data frame and write to tsv.
     df = pd.DataFrame(dars)
     df = df.rename(columns={
-        "Abbreviation": "consent_group"
+        "Abbreviation": "consent_group",
+        "Request Date": "request_date",
+        "Last Renewal Date": "last_renewal_date",
     })
 
     # Replace newlines in study with spaces.

parse_dbgap_application.wdl

@@ -37,7 +37,7 @@ task extract_dars {
     }
 
     runtime {
-        docker: "uwgac/parse-dbgap-application:0.1"
+        docker: "uwgac/parse-dbgap-application:0.2"
     }
 }
 
@@ -54,6 +54,6 @@ task render_report {
         File dar_report = "dar_report.html"
     }
     runtime {
-        docker: "uwgac/parse-dbgap-application:0.1"
+        docker: "uwgac/parse-dbgap-application:0.2"
     }
 }