Sunday, April 25, 2010

University ranking on Bioinformatics [ 생명정보학 대학 랭킹 ]


I ranked universities according to the number of publications in journals on bioinformatics. The journals concerned for this task are Bioinformatics, BMC Bioinformatics and Plos computational biology. The journal selection criteria might be controversial since articles on bioinformatics have been published nearly all journals on 'biology', including top science journal Nature or Science. However, I selected the three journals for further analysis since it requires additional mandatory validation for articles published in the other journals to identify whether a given article is on bioinformatics or not( I used automatic selection and counting approach for this task). Moreover, the selected three journals are top ranked journals in the field of bioinformatics ( according to ISI impact factor ) so that ranking according to the publication numbers in these journals is not depend only on number of publications but the number of 'high' quality publications.

METHOD

  • Search journal : I used CPAN perl module 'WWW::Search::PubMed' to retrieve information of papers published in three journals. To retieve Affiliation information, I added lines of codes.
  • Control variation of university name : Majority of papers used standard official name to specify each university. Howevr, some of papers used university name which is subtly different from the standard one. Since it requires mandatory validation, I didn't consider these variations. I only consider universities are the same if they used exactly the same name. However, I controled terms for 'University of California' . There are branches of universities for 'University of California', such as UC san diego, UC los angeles , UC berkely, etc. And generally, the university name is divided by comma (, ) which is a seperator in affiliation line into two parts, 'University of California' and city name such as 'Los Angeles', 'San diego' , etc. So I combined university name and the city name if 'University of California' is stated alone followed by comma without city name.

Overall ranking
( Top 50 )

#colum 1: Ranking
#colum 2: Number of publications
#colum 3: University name
( Non US universities are orange colored )

1 108 UNIVERSITY OF CALIFORNIA SAN DIEGO
2 88 STANFORD UNIVERSITY
3 72 COLUMBIA UNIVERSITY
3 72 UNIVERSITY OF WASHINGTON
3 72 UNIVERSITY OF MANCHESTER
6 71 UNIVERSITY OF MICHIGAN
7 69 YALE UNIVERSITY
8 63 UNIVERSITY OF CALIFORNIA BERKELEY
8 63 UNIVERSITY OF CAMBRIDGE
10 60 WASHINGTON UNIVERSITY
11 59 UNIVERSITY OF OXFORD
12 58 TEL AVIV UNIVERSITY
12 58 UNIVERSITY OF TOKYO
14 56 NATIONAL UNIVERSITY OF SINGAPORE
15 53 PRINCETON UNIVERSITY
16 52 UNIVERSITY COLLEGE LONDON
17 48 UNIVERSITY OF MINNESOTA
17 48 UNIVERSITY OF QUEENSLAND
19 47 IOWA STATE UNIVERSITY
20 45 BOSTON UNIVERSITY
21 44 UNIVERSITY OF TORONTO
22 42 UNIVERSITY OF CALIFORNIA IRVINE
22 42 UNIVERSITY OF PITTSBURGH
24 41 CORNELL UNIVERSITY
24 41 KYOTO UNIVERSITY
24 41 UNIVERSITY OF SOUTHERN CALIFORNIA
27 40 UNIVERSITY OF CALIFORNIA SAN FRANCISCO
28 39 KAIST ( Korea Advanced Institute of Science & Technology )
28 39 TEXAS A&M UNIVERSITY
28 39 UNIVERSITY OF BRITISH COLUMBIA
31 38 JOHNS HOPKINS UNIVERSITY
32 37 HARVARD UNIVERSITY
32 37 UNIVERSITY OF CALIFORNIA DAVIS
34 36 UNIVERSITY OF PENNSYLVANIA
35 35 CARNEGIE MELLON UNIVERSITY
35 35 UPPSALA UNIVERSITY
35 35 UNIVERSITY OF MARYLAND
38 33 BIELEFELD UNIVERSITY
39 31 UNIVERSITY OF CALIFORNIA SANTA CRUZ
40 30 PEKING UNIVERSITY
40 30 UNIVERSITY OF TEXAS SOUTHWESTERN MEDICAL CENTER
40 30 UNIVERSITY OF GEORGIA
43 29 INDIANA UNIVERSITY
43 29 UNIVERSITY OF GLASGOW
43 29 UNIVERSITY OF CALIFORNIA LOS ANGELES
43 29 TSINGHUA UNIVERSITY
47 28 OHIO STATE UNIVERSITY
48 27 Seoul National University
49 26 TECHNICAL UNIVERSITY OF DENMARK
49 26 PENNSYLVANIA STATE UNIVERSITY



Rising universities

After HGP, bioinformatics recogenized as one of the most important field of genome revolution. Universities that did not specially drive their effort on bioinformatics before the time had started bioinformatics research with large investment. Therfore, the overall ranking might have undergone lots of changes.

To find these variation, I ranked again according to relative rank increase( rank_after_2001 - rank_before_2005 / rank_before_2001).
Range of publication time is divided into two duration, before 2001 and after 2005.
Universities without any publication before 2001 were not considered for this analysis.


[ Top 10 raising university]

#colum 1: Ranking
#colum 2: University name
#colum 3: relative rank difference
#colum 4: ranking for 1998-2000
#colum 5: ranking for 2006-2010

1.UNIVERSITY OF CALIFORNIA SAN DIEGO 0.93 15 1
2.UNIVERSITY OF WASHINGTON 0.92 39 3
3.STANFORD UNIVERSITY 0.66 6 2
4.TEL AVIV UNIVERSITY 0.56 39 17
4.UNIVERSITY OF QUEENSLAND 0.56 39 17
6.YALE UNIVERSITY 0.5 6 3
7.IOWA STATE UNIVERSITY 0.41025641025641 39 23
8.UNIVERSITY OF SOUTHERN CALIFORNIA 0.35 39 25
8.UNIVERSITY OF TORONTO 0.35 39 25
8.KYOTO UNIVERSITY 0.35 39 25


Thoughts on the ranking


1. Can I say that high ranked universities are better than universities in low ranked?
  • University is different from research institution. I think university ranking should consider not only research performance, but also how successfully they educate student. This ranking is based only on number of publications so that it only consider research performance. Therefore, this ranking should not be considered to reveal overall value of universities.

2. How does UCSD excel the other top schools?
  • UCSD is ranked around 7-10 when the ranking is based only on 'Bioinformatics' or 'BMC bioinformatics'. However, UCSD is top ranked with overwelming number of publications when it comes to 'Plos computational biology' journal. Since Plos family journals were initiated with tight relationship with faculty members at UCSD, they have willingness to publish papers in Plos journal. Moreover, Philip Bourne, a professor at UCSD in the field of bioinformatics, has been contributed a series of editorials, ' Ten simple rule ' advise series, in Plos Computational biology, which are not research papers, but counted as equally as research papers to be considered for ranking.
3. Some most prominent universities such as Harvard ranked lower compared to others, are they really bad on Bioinformatics or computational biology?
  • As I described in the beginning, this ranking is only based on top three journals in the field of bioinformatics or computatinoal biology. If the other journals were considered, the ranking would be changed a lot.
4. USA vs. the other countries
  • In the top 10 list, only Cambridge and Manchester are universities out of USA. In top 20, there are 8 and 12 in top 30. Regardless of drawbacks of ranking criteria, it's ture that USA is a leading country in this field.
  • List of countries other than USA in top 30 ranking are Japan, Canada, Israel, Korea, Singapore and Austrailia. 3 Aisan countries have 4 universities ranked within top 30.