Recently I had the unusual experience of being an advisor for hire. My client was a clinical scientist specialized in molecular biology. One of his post-docs had started a project that required bioinformatics expertise. She already had a computer science background and had taken some training in transcriptomics data analysis but she was inexperienced in this methodology. The PI could not assess her work and guide her. So I became her advisor.
We met once every week or two. I helped her getting set up with the High Performance Computing environment of Yale University and the intricacies of some Unix commands. She needed to combine data from multiple files into a master table, then remove the low quality or inconsistent data. We discussed the analytic strategy, decided on the best way to validate the data. I was a combination technical consultant and academic advisor. Our meetings were partly trouble-shooting and partly planning sessions for what she needed to do.
I was glad to be able to assist the postdoc and enable her to complete the analysis. But we were determined not to take the data at face value and our effort paid off. As it turned out, the data did not validate, the genotype was not consistent with the genetic crossings that were performend. This meant that the data was not usable and the project had to be terminated. Our consolation was that it is better to fail early than rush to publish and be embarrassed later.