Analyzing sleep in Ethoscope based on beam crosses

[mz27ethio]

Greetings,
Me and my lab mates have been using the Ethoscope to study sleep in several Drosophila mutants. As it can be expected, the sleep data we analyzed based on the DAM system is different from that of the Ethoscope. Thus, we wanted to analyze sleep in Ethocope in DAM fashion to better understand the difference. Hence, we wanted to analyze sleep not in terms of not-moving for 5 mins (300 sec), but in terms of not-crossing the beam for 5 min. As it can be seen in the raw Ethoscope data (image attached), beam_crosses information is provided in 10 sec interval alongside other parameters. Since I am new to R, I don't know efficient ways to make this analysis. However, I just tried the following double loop to compute a data table for sleep that is based on beam crosses:

initialize the data table; dt1 is my raw data from the Ethoscope;

         sleep_check <- data.table((dt1[1,"id"]), (dt1[1, "t"]), asleep=paste("FALSE"))

##since the first row is initialized start the loop from second row

       for (row in 2:nrow(dt1)){
            id_name=dt1[row, "id"]
            time=dt1[row, "t"]
             count=row

         dt_tempo<- data.table(id_name, time, asleep="TRUE")	

             while (count<row+30 && count<nrow(dt1)){
                 ##loop for 30 data points to analyze beam_crosses for 300 second

	         beam_crossed=dt1[count,"beam_crosses"]
	         count=count+1	
				
                 if (beam_crossed>0)
                       {
                           dt_tempo<- data.table(id_name, time, asleep=paste("FALSE"))
	                   break
                       }							
               }	
         sleep_check <- rbind(sleep_check, dt_tempo)	
      }

However, since my raw data has more than 3,000,000 rows, the computation is running for a very long time, like more than 20 hours. I wanted to ask if you can show me a better way of approaching this problem.

Thanks very much

[pepelisu]

To compare results of ethoscopes and DAM monitors, I think the best option is to use the .txt file that should be in the same folder as the .db, that file contains the data but ready to use in your DAM analysis workflow.

Apart from that I suggest to use the rethomics package to analyze the .db files and I would recommend to take a look to this tutorial https://rethomics.github.io/ and to the data.table documentation data.table github page

[qgeissmann]

Hi,
Yes, I agree with @pepelisu , it is a job for rethomics.
To give you a hint, you need NO explicit for loops in your code. You can do everything in a vectorial fashion.

1. Define the middle of the tube as zero, on your coordinate axis (- on the left, plus on the right)
2. Find all changes of sign + -> - and - -> + , set them to 1, the rest to 0 (arbitrary)
3. Use something like Run Length Encoding (rethomics bout analysis) to characterise onset, duration/... of 0 bouts

Closing for now, happy to help more in the future :),

[mz27ethio]

@qgeissmann and all, thanks very much for the help. I will go ahead and try it. In the mean time though I wanted to ask one thing. In the raw Ethoscope Data, there is a column that says "beam_crosses." Can you please clarify what this is? My assumption is that it is the number of times the fly crosses the middle of the tube in the 10 second interval.