July 10th, 2016
This week we worked on importing the primer plate barcode data into the RECCS mapping file using Rstudio, making an OTU table from the sequencing data using Unix, performing a Bray-Curtis and PCOA statistical analysis on all the plate data, and performing a Kruskal Wallis statistical analysis of the PMA samples data. During the communication workshop I made a quick concept map and gave an informal presentation.
The initial set of data that we copied over from the sequencing center was cause for some major problems. Once Tess and I began to work with the data to format it into a usable OTU table, hardly any of the barcodes matched up with the barcodes in the sample, meaning barely any of the sequences were kept for data analysis. This was a huge red flag and at first I thought that it was all due to my own experimental error. Perhaps the barcodes didn’t attach to the samples like they were supposed to. Another set of researchers had sent off their PCR product to be sequenced mixed in with ours which had their own primer barcodes on them, but they had the same problem with the sequencing file: no data to work with. Even though this was bad, it was easy to tell that something wasn’t right since everyone’s data was slim from the sequencing file. We eventually determined that the sequencing file we had copied over was wrong, found the correct one, repeated the conversion process and obtained an OTU table full of tons of data. This was such a relief. I’m very happy that I have a lot of data to work with for analysis!!
When we were working with the incorrect sequencing data and hadn’t determined that it was the wrong file yet; I was feeling incredibly disappointed. I felt like all of the work that I had done in the past month was all for nothing. Tess gave me some great advice when she noticed this. She told me that I shouldn’t take the results personally, because circumstances like this happen in science all of the time. I asked her if something like this had ever happened to her with some of her own experiments and she said that it had several times. Sometimes she had to redo her experiment. It definitely gave me some relief to know that just because my data looked bad, I shouldn’t blame myself for the results. A multitude of different issues could have been the reason for the results I was seeing. In the end it worked out! The file was wrong and I was grateful to see the real data. I certainly felt accomplished that the data I managed to collect was abundant! I’ve definitely learned that it’s critical not to immediately blame myself or my ability if something goes wrong in the experiment.
This week has gone well overall. I feel quite proud of myself for catching on to the use of R language. It’s definitely helpful when you keep track of all the commands in an R script so you can reference them later on when you want to repeat the same function with more data. It is an extremely gratifying feeling to work with a massive file of data and to filter it down into a nice, readable figure or graph. I’m looking forward to continuing my data analysis!
I remember you talking about this in your informal presentation. Good thing there was a way to track down the problem. Making progress and learning that all sorts of things can happen to your experiments!