Python Programming for Biology: Bioinformatics and Beyond

Reading PubMed XML files

PubMed

3

is a search engine that lets you access the MEDLINE database of citations for

articles  in  the  life  sciences.  You  can  download  the  citations  in  various  formats.  In  this

section we consider the XML format.

As  with  all  XML  formats,  although  you  can  escape  from  the  pain  of  parsing  using

ElementTree,  you  still  have  to  understand  what  the  schema  (or  ‘data  model’)  is.  The

schema  for  PubMed  XML  is  defined  by  a  DTD  (Document  Type  Definition),  although

reading this is not very enjoyable. And as with all schemas it is quite possible that it will

change in future, so application code can break. Here we use the DTD that was valid on 1

January  2009.  We  will  show  how  to  extract  and  print  the  journal  year  and  title  and  the

article title and abstract, from a collection of PubMed XML files (which, for example, can

be downloaded from the PubMed website).

The  root  object  has  the  tag  ‘PubmedArticleSet’  and  underneath  that  are  one  or  more

children  with  the  tag  ‘PubmedArticle’,  although  here  we  will  just  look  at  the  first  child.

Underneath  that,  there  is  either  a  child  with  tag  ‘NCBIArticle’  or  ‘MedlineCitation’  and

we  will  assume  the  latter.  Continuing  down  the  hierarchy  we  eventually  get  to  the

information we want:

def printPubmedAbstracts(xmlFiles):

for xmlFile in xmlFiles:

tree = ElementTree.parse(xmlFile)

root = tree.getroot()

citationElem = root.find('PubmedArticle/MedlineCitation')

pmid = citationElem.findtext('PMID')

articleElem = citationElem.find('Article')

journalElem = articleElem.find('Journal')

journalTitle = journalElem.findtext('Title')

journalYear = journalElem.findtext('JournalIssue/PubDate/Year')

articleTitle = articleElem.findtext('ArticleTitle')

articleAbstract = articleElem.findtext('Abstract/AbstractText')

print('PMID = %s' % pmid)

print('journalYear = %s' % journalYear)

print('journalTitle = "%s"' % journalTitle)

print('articleTitle = "%s"' % articleTitle)

---

print('articleAbstract = "%s"' % articleAbstract)

print('')

The PMID is the PubMed ID of the citation. With variants of this code you could create

your own kind of short summary of MEDLINE citations.