Citation network analysis of retractions in molecular biology field

Based on the extracted information of retracted papers from the Retraction Watch Database and the citation information of these papers from the Web of Science, we uncovered the complex relationships of retracted papers in the molecular biology domain via a citation network. The basic characteristics (i.e., time and spatial patterns, reasons, publishers) of the retracted articles were studied. Citation network analysis, including community detection and text analysis, was carried out. Our main findings are as follows: (1) The overall number of retractions in this field has been increasing over time, and these retractions have been mainly in China and the USA. (2) Most retracted papers were for both “scientific error” and “misconduct” reasons. Among the 13 reasons given, errors in the data and images accounted for the largest proportion. (3) Community structure is obvious in the citation network we constructed. In communities with five or more nodes, the average self-citation rate account for 76%. In the three largest communities 1, 2, and 3, the self-citation rate are respectively 99%, 100% and 77%. In community 6, the self-catition rate is 17%. Other papers from different teams were published in the Journal of Cellular Biochemistry (4 papers). Tumor Biology (3 papers) or Febs Letters (1 paper). The self-citation rate of community 5 is 60.00%. Most papers are from Alfredo Fusco’s team, and other ten papers are almost published in PLoS ONE. (4) The coupling relationship between citing-cited retraction reasons was revealed. Retractions and their citations were more likely to be retracted for the same reason. Most of the citing-cited papers from paper mills were published by the same publisher and even the same journal. (5) PI3K (an enzyme), WNT (a protein) and lncRNAs have recently been the major topics of retractions.