Retraction Watch

Tracking retractions as a window into the scientific process

Sampling error, flawed analysis, and miscalculation trigger Molecular Cell retraction

with 4 comments

molecular cell 14Guest post by Jean Hazel Mendoza

A group of researchers from France has retracted a 2013 paper from Molecular Cell after realizing that their analyses of microscopy images were flawed.

Here’s the notice for “RecA-Promoted, RecFOR-Independent Progressive Disassembly of Replisomes Stalled by Helicase Inactivation:”

This article has been retracted: please see Elsevier Policy on Article Withdrawal (

This article has been retracted at the request of the Authors. In our paper, a collaboration between the Allemand and Michel labs, single-molecule microscopy experiments were performed to describe the fate of blocked replication in an E. coli helicase mutant. The image analyses were performed by the first author of the paper. Our recent attempts to repeat these analyses have failed, and we now realize that several issues invalidate the original analyses, including sampling errors, skewed statistical analysis, and miscalculations. We therefore wish to retract this paper. We apologize to the scientific community for any loss of time and resources caused by this publication.

The study has been cited five times, according to Thomson Scientific’s Web of Knowledge. Its abstract reads:

In all organisms, replication impairment is a recognized source of genomic instability, raising an increasing interest in the fate of inactivated replication forks. We used Escherichia coli strains with a temperature-inactivated replicative helicase (DnaB) and in vivo single-molecule microscopy to quantify the detailed molecular processing of stalled replication forks. After helicase inactivation, RecA binds to blocked replication forks and is essential for the rapid release of hPol III. The entire holoenzyme is disrupted little by little, with some components lost in few minutes, while others are stable in 70% of cells for at least 1 hr. Although replisome dissociation is delayed in a recA mutant, it is not affected by RecF or RecO inactivation. RecFOR are required for full RecA filaments formation, and we propose that polymerase clearance can be catalyzed by short, RecFOR-independent RecA filaments. Our results identify a function for the universally conserved, central recombination protein RecA.

The authors report that their microscopy procedures and image analyses were based on a study they had published 2012 in Science, which itself has been cited 22 times, according to Thomson Scientific’s Web of Knowledge:

The microscopy set up is as in Lia et al. (2012) and is described in Supplemental Experimental Procedures. All image analyses were performed using the software ImageJ as described previously (Lia et al., 2012) (see also Supplemental Experimental Procedures and Figure S2). At least 80 cells (most often 150–350) were used for each determination.

When and how issues with the images arose in the Molecular Cell paper remains unclear. Lead authors Bénédicte Michel and Jean-François Allemand have yet to reply to requests for more details (first author Giuseppe Lia’s Centre de GénétiqueMoléculaire email bounced). We’ll update this post with anything we learn.

Written by Ivan Oransky

May 23rd, 2014 at 9:30 am

  • Stewart May 23, 2014 at 3:53 pm

    An interesting one.

    “The image analyses were performed by the first author of the paper. Our recent attempts to repeat these analyses have failed, and we now realize that several issues invalidate the original analyses, including sampling errors, skewed statistical analysis, and miscalculations. ”

    The only errors were from one author? Surely it cannot have been so serious as to warrant a retraction. Maybe the first author was very skilled and the others were unable to reproduce the excellent work due to lack of ability. Perhaps the failed reproduction was attempted in the collaborators laboratory (who may have been, erm, misinformed of certain technical aspects of the methodology by their “collaborators”). This is not uncommon. Indeed, it is very common.

    Does this put doubt on large, multiscale national or international scientific collaborations where politics/gamesmanship overrule best scientific practice? I know of several such collaborations, and none of them run smoothly, quite the opposite.

    I think this retraction is untimely. I doubt such simple, honest alleged errors (afterall, we are all human) would alter the discussion, message or conclusions of the paper.

    Surely the “authors” should have requested another independent laboratory to attempt to reproduce the work prior to retraction.

  • ferniglab May 28, 2014 at 12:49 pm

    Exact replication may be difficult, if the ‘microscope” is in fact itself and optical experiment – this is often the case in single molecule microscopy. One consequence is that it is easy to be misled by data, e.g., due to systematic biases in the data which arise from the measurement process, but which only the microscope builder is aware of. The builder may have moved on. So it may be that the first author used tools produced by a previous lab member and didn’t realise the skews inherent in some aspects of the data. These skews passed over the heads of the co-authors, result retraction. However, without any detail it is not possible to figure out the underlying cause.

  • Jil71 May 28, 2014 at 3:13 pm

    It seems that the first author has resigned from his CNRS position. This might have somehow curtailed a deeper enquiry / lenghty re-analysis of the data…

  • fernando pessoa December 20, 2014 at 6:26 pm

    Another 2014 retraction by the same authors.

  • Post a comment

    Threaded commenting powered by interconnect/it code.