USV Analysis - In depth

[hagenaue]

This is a little mini-project to do a more in depth analysis of the F53 and F56 generation ultrasonic vocalization (USV) data. Ultrasonic vocalizations from rodents can signal affective state (e.g., distress, joy/social approach - see background reading below). These USVs were recorded during the social interaction test, and thus reflect the state of two different animals - the bHR/bLR animals within our experiment and the "stimulus" animals that they were interacting with.

Currently, the USV data were quantified using two broad categories: "22 kHz" calls, which are typically viewed as distress calls, and "50 kHz" calls, which are typically viewed as reflecting positive affect of some sort (joy, social approach). However, the "50 kHz" calls may actually reflect a broad range of frequencies (32-96 kHz) and could potentially be further subdivided into many other categories. It would be nice to have a broad overview of the call characteristics (frequency distributions, duration distributions) and whether those characteristics seem to change with treatment (bred line, environmental enrichment, social defeat).

The detailed USV recording files (including the peak & duration for each USV call made during the social interaction test) for each animal are located on MBox in: All Files->UROP-2019->AngelasExperiment_Data->USVs_ForHistograms https://umich.box.com/s/c1jyefbovswjc3v1sxwt8s6mu50ex8p7

Unfortunately, the F53 and F56 generations were recorded separately using slightly different recording settings (gain, etc), so we will need to keep the results from the two generations separate while performing our analysis.

Our two goals are to make histograms illustrating the peak frequency and duration for all calls made by: 1) All animals from the F53 generation (I already did the F56 generation - I've included their code and histograms) 2) All animals from each treatment group in the F53 generation and in the F56 generation.

I've included information about each of the processing steps in comments below.

[hagenaue]

Here is some background reading about ultrasonic vocalizations in rodents and how they communicate affective state: Preview of “Ultrasonic vocalizations (USV) as a tool for research on emotion and motivation in rodents”.pdf Types Functions of USVs.pdf

[hagenaue]

Goal 1: Make histograms illustrating the peak frequency and duration for all calls made by all animals from the F53 generation.

I already did the F56 generation - I've included the code for the F56 analysis here: https://github.com/hagenaue/Angela_HRLR_StressEnrichData/blob/master/11_DetailedUSVAnalysis.R

And the histogram output for the F56 generation is here: https://umich.box.com/s/yn4qus1k3f9iozwctuk2adczbkafxg6w https://umich.box.com/s/mkw73gbhyps7rabsfee1dndk8shaiswm

To perform the same analysis on the F53 data is going to require organizing some files and adapting the code a little. In particular, Angela currently has all of the F53 USV data in three folders labeled Run 1, Run 2, and Run 3: https://umich.box.com/s/k31q8435cqkmo0di6i33ipqq2noufceq

If you want to recycle my F56 code, all of the F53 animals will need to be placed in the same folder while retaining their Run information. There are two easy ways to go about this: 1) Rename all of the F53 files with the run included in the title, e.g. "Run1_Recording39" and then drop all of the F53 files into a new F53_all folder. 2) Rename all of the F53 files with their actual animal ID found in the decoder file: https://umich.box.com/s/pb6bkmbdlu4asttyj5pxjfliss730se7 (This may also help with the next task...maybe).

... and then tweak the code document to fit the F53 analysis.

[hagenaue]

Goal 2: To make histograms illustrating the peak frequency and duration for all calls made by all animals from each treatment group in the F53 generation and in the F56 generation.

This task is substantially more of a pain in the butt.

1) The animal IDs & treatment groups for each of the recordings is found in this decoder file: https://umich.box.com/s/pb6bkmbdlu4asttyj5pxjfliss730se7 ... but, as we have already learned from other analyses, Angela often doesn't use the exact same formatting for her Animal IDs in different documents. So before doing anything, we may need to double-check whether each of the Animal IDs in the decoder document is named using the same Animal ID formatting as our other master spreadsheet.

2) When reading in each file, we will want to associate the file with its Animal ID and treatment group. This could potentially be done in two ways:

• A) You could rename each file with its Animal ID (instead of "Recording 39"), and then when reading in the data from the file add a new column called "file name" to the data frame. Later, this data frame can be joined with the decoder to obtain information about treatment group.
• B) You can keep the original file name. Then, when reading in the data, you add a new column called "file name" to the data frame, and then join this data frame with the decoder to obtain information about both Animal ID and treatment group. If you take this strategy, you will need to make sure that the file name information in the decoder exactly matches the actual file names.

3) Then, before making the histograms, subset the data frame by treatment group and output a different histogram for each treatment group, keeping the F53 and F56 generation data separate.

[liamctf]

I uploaded the updated histograms (which have been parsed by bred line and treatment group) to the box folder