Category: Referencing & Document Control

 

1. Don’t ever use loose bits of paper. They get lost.
2. Date everything, including your electronic files and rough calculations.
3. Write down any changes you make to procedures, so that if the outcome changes you have some possible explanations.

 
I did some experiments where I had to add about 10 different inhibitors to cells and work out the best concentrations for each. Each time I scrawled down the values for each inhibitor without labels. I knew at the time which was which, but a few weeks after when I came to analyse it, this information had long since departed me. It was stressful and time consuming to work out which concentrations I had used for which drug at which time.

 

1. In Excel, keep a copy of all your graphs on a separate sheet (not a separate document), which should be labelled graphs. Then you can instantly see if a spreadsheet contains good data or not.
2. Label all your graphs with the maximum amount of information. Don’t leave the axis as x or y or % Change. They are meaningless. Write exactly what you measured in the axes.
3. Try as much as possible to utilise (and title) different sheets (again NOT different documents) for different tables of data. If you have to scroll through thousands of lines of data with other tables beneath and to the side, it quickly becomes uninterpretable.

 
My bane was always the unlabelled graph. I would even show them to my supervisor and he’d just look at me blankly and say, “What does it mean?” At the time it was obvious to me as I’d just done it, but he – and myself in the future – had no idea. It’s so quick to label an axis, and it takes so long to interpret one that isn’t labelled. When I wrote my thesis I discovered this, but it was too late by then.

 

1. On your computer, divide your research up into folders for each project, labelling each folder with the project acronym. Store all your data and protocols relevant to that project in that folder.
2. For each replicate add a number to the acronym, and make folders for each replicate within the project folder. Make sure you write the outcome and the reason for doing the replicate on the results. You could even add a one letter identifier to the file name so you can quickly see whether it contains good data, “G” or bad data, “B”.
3. Date everything, as this provides a failsafe for linking bits of work on different computers, or from computers to lab and notebooks.
4. Copy all data off communal machines as soon as possible.

 
When it came to writing my thesis, I spent ages trawling through data with incomprehensible file names. Many of my graphs were already in journal papers, but some of them needed subtle changes and others had disappeared from the stats software, so they needed to be done again. Some of these sets of data were almost three years old, and I’d long forgotten what it all meant. If I’d put it all in separate folders with succinct file names it would have taken a fraction of the time to remake those graphs.

 

1. Think about your labels for more than the minimum number of seconds. If you have just written a few words, then you will probably have missed something.
2. Date everything. If all else fails, the date may still save you.
3. When you date your files on computer, write them backwards, starting with the year then the month and then the day, so the computer will automatically put your files in chronological order.
4. Computers have lots of memory, so write accordingly long file names. You can never have too much information immediately visible.
5. Even if you are only using something for a few minutes, it is still worth labelling. You could drop things or mix them up, and the briefest loss of concentration could otherwise leave you unable to continue.
6. Keep all the datasheets of the products you order in a draw, separate from everything else.
7. For papers, copy and paste the whole title into the Save As.

 

1. If the tube is too small to label properly, put it in a bag and label that.
2. If the material is difficult to write on, write on a piece of paper and then stick that on.
3. Be careful you don’t label liquid containing tubes with pens that will dissolve in the solvent.

 
I once did a proteomics study where I labelled each tube in a red pen only to find that the alcohol in the tubes had spilled through the holes in the tops and leached the pen back in with it. The result was a series of tubes with varying degrees of pink in them, and the uncertainty as to whether the degree of pinkness had affected the results.

 

1. Every paper you want to read or reference later should be saved as a PDF on your computer before you add it to a library such as Readcube, Mendeley or Endnote. If your other libraries get deleted or overwritten – which can happen – then you at least have all the PDFs.
2. Save the PDFs with the full title of the article, not just an abbreviation or vague description, and organise your folders by topic so you can see what papers you have read in each area of your research.
3. You can always print them out to read, but if you intend to refer to them later a virtual copy is advisory. One exception is when there are bits of protocols that you might want to cut out and stick in your lab book.

 
I printed out most of the papers I wanted to read, and most never got read, which was pretty wasteful. Beyond that, the stacks of papers for my thesis were stored in piles in my room, numbered so that I could just write (ref 10) and not interrupt the flow of writing. When it came to putting in all the references, I thought I could just sit down in front of Breaking Bad and plug them in.

WRONG.

I still don’t know how, but several of them had gone missing, and other times I had written down the wrong reference number in my thesis. Referencing took a whole lot longer than it would have if I had both used electronic documents and referenced as I wrote.