Box 2. When outsourcing gets ugly—a case study
From the following article
The power and pitfalls of outsourcing
Franz B Pichler & Susan J Turner
Nature Biotechnology 25, 1093 - 1096 (2007)
doi:10.1038/nbt1007-1093
A quotation is provided by a CTP for sequencing of a bacterial genome. The service includes preparation of genomic DNA from the client-supplied culture, construction of clone libraries, shotgun sequencing to 8
coverage and data assembly. Assurances of quality control are also provided. The quotation is accepted subject to inclusion of three milestone checkpoints at which time the client has the opportunity to review the data and terminate the program if unsatisfactory. Milestone 1 is reached. The culture has been provided to the CTP, libraries have been constructed and screened for bias. Results indicate satisfactory performance. Both parties agree to progress to the next stage. Milestone 2 is reached. Shotgun sequencing is completed to 1
coverage and partial assembly completed. The CTP recommends progression to the next stage (completion of sequencing), but a statistical analysis by the client raises concerns about the unusually high number of contigs given the genome size. The CTP reviews data and concludes that the results are within acceptable norms. Further analysis by the client reveals that the sequence data are contaminated with reads from another organism. Questions about the source of the contamination and the quality control measures undertaken by the CTP cannot be resolved and the parties agree to terminate the contract. Sufficient time and funds remain for the researchers to obtain services from another provider.
A key point in this case was that the contamination issue was only detected because of the client's vigilance in carrying out their own assessment of the data. Detection of the problem relied on the availability of in-house expertise in genome sequence assembly and analysis. This case study highlights two important points. First, milestones and checkpoints relating to quality and performance are a useful way of ensuring a degree of client-control over large or complex projects. Even so, these measures are useful only if the client has sufficient knowledge to provide a critical assessment of performance. A certain degree of in-house knowledge is still required to ensure that experimental results are within accepted standards.
