5 Make DAM Output

Steps to convert iLAM output into Trikinetics DAM format for locomotor activity analyses

Read in iLAM output from image_analysis and parse into a dataframe containing the size, centroid location, and origin (time, pi) of all identified blobs (by_change) with parse_movements()

out_file_name = "Pgreeni"
start_photophase = 5 #project-specific, time that lights turn on
end_photophase = 21 #project-specific, time that dark starts

#parse movements into workable dfs: all movements (by_change) and
by_change_Pg <- parse_movements(file_mvmnts = paste0(out_file_name,".csv"),
                             start_photophase = start_photophase,
                             end_photophase = end_photophase)

Table 5.1: iLAM by_change data frame
…1	s	x	y	pi	time	ID	sex	genus	species	treatment
1	2000000	0	0	mothra03	2022-06-24 21:00:00	99.9%_s10g15	male	photinus	greeni	D
2	2799	1256	35	mothra03	2022-06-24 21:02:00	99.9%_s10g15	male	photinus	greeni	D
3	2189	150	466	mothra03	2022-06-24 21:02:00	99.9%_s10g15	male	photinus	greeni	D
4	731	1282	40	mothra03	2022-06-24 21:04:00	99.9%_s10g15	male	photinus	greeni	D
5	442	1311	51	mothra03	2022-06-24 21:04:00	99.9%_s10g15	male	photinus	greeni	D

Collapse all identified blobs (within by_change_[]) into a dataframe containing the sum of blobs and number of blobs/movements observed per frame/timepoint (by_frame_[])

by_frame_Pg <- 
  by_change_Pg %>% ungroup() %>% 
  group_by(pi, ID, time, treatment) %>%
  dplyr::summarize(n = length(s[!is.na(s)]), #number of blobs of size (s) != NA
            s = sum(s, na.rm = TRUE)) %>% #sum of blob sizes, NA removed
  dplyr::mutate(n = ifelse(s == 0, 0, n)) %>% #if sum of blobs=0, then n<-0 (otherwise it'd be 1)
  distinct(pi, ID, time, treatment, n, s) #sanity check to remove any duplicates

Filter movements captured from rare, noisy images.

Noisy values were initially assigned an arbitrary value (2,000,000) by find_movements(). This step changes 2,000,000 with NA and then replaces with average value of the preceding and succeeding time points. These rare cases typically only occur at the image following a light-dark transition.
- For example, across a 7-day experiment (5040 images), only 0.4% image comparisons were assigned an arbitrary value.

library(zoo)

by_frame_Pg <-
  by_frame_Pg %>% ungroup() %>%
   mutate(s = replace(s, s==2000000, NA),
          n = replace(n, is.na(s), NA))

by_frame_Pg <-
  by_frame_Pg %>% group_by(pi) %>%
  mutate(s = round((na.locf0(s, fromLast = TRUE) + na.locf0(s, fromLast = FALSE))/2,0),
         n = round((na.locf0(n, fromLast = TRUE) + na.locf0(n, fromLast = FALSE))/2,0)) %>% ungroup()

by_frame_Pg <-
  by_frame_Pg %>% ungroup() %>%
   mutate(s = replace(s, is.na(s), 0),
          n = replace(n, is.na(n), 0))

dplyr::filter(by_frame_Pg, n>0) %>% group_by(pi) %>%
  dplyr::summarise(blob_size = mean((s/n), na.rm=TRUE)) #output the average blob size

Table 5.2: iLAM by_frame data frame
…1	pi	ID	time	treatment	n	s
1	mothra01	99.9%_s10g15	2022-06-24 21:00:00	D	0	0
2	mothra01	99.9%_s10g15	2022-06-24 21:02:00	D	4	4084
3	mothra01	99.9%_s10g15	2022-06-24 21:04:00	D	2	2684
4	mothra01	99.9%_s10g15	2022-06-24 21:06:00	D	5	8803
5	mothra01	99.9%_s10g15	2022-06-24 21:08:00	D	9	18515

Use iLAM wrapper make_dam_file() to convert by_frame_[] into a DAM-like output and save as a DAM file with summed blob sizes/total movement (in pixels) per timepoint

ilam_Pg = make_dam_file(by_frame_Pg, variable_name = "s") 

#output a tab-delimited file containing 
write_tsv(ilam_Pg, "ilam_Pg.tsv", col_names = F)
write_tsv(ilam_Pg, "Monitor42.txt", col_names = F) #some analysis programs require Monitor##.txt name

Table 5.3: iLAM by_frame converted to DAM format
X1	X2	X3	X4	X6	X8	X11	X12	X13
1	24 Jun 22	21:00:00	1	42	Ct	0	0	0
2	24 Jun 22	21:02:00	1	42	Ct	4084	1668	4988
3	24 Jun 22	21:04:00	1	42	Ct	2684	2981	3666
4	24 Jun 22	21:06:00	1	42	Ct	8803	22280	4058
5	24 Jun 22	21:08:00	1	42	Ct	18515	17325	5093

Use iLAM wrapper make_dam_file() to convert by_frame_[] into a DAM-like output and save as a DAM file with the number of blobs/movements per timepoint

#parse movements into workable dfs: all movements (by_change) and
by_change_Pg <- parse_movements(file_mvmnts = paste0(out_file_name,".csv"),
                                start_photophase = start_photophase,
                                end_photophase = end_photophase)

by_frame_Pg = filter_nblobs(by_change_Pg) #filter out small nblobs using iLAM wrapper filter_nblobs()
ilam_Pg = make_dam_file(by_frame_Pg, variable_name = "n")

write_tsv(ilam_Pg, "ilam_Pg_n.tsv", col_names = F)
write_tsv(ilam_Pg, "Monitor43.txt", col_names = F) #some analysis programs require Monitor##.txt name

Table 5.4: iLAM by_frame converted to DAM format
X1	X2	X3	X4	X6	X8	X11	X12	X13
1	24 Jun 22	21:00:00	1	42	Ct	0	0	0
2	24 Jun 22	21:02:00	1	42	Ct	2	0	2
3	24 Jun 22	21:04:00	1	42	Ct	2	1	2
4	24 Jun 22	21:06:00	1	42	Ct	5	3	3
5	24 Jun 22	21:08:00	1	42	Ct	8	6	4