natfeedback_code_eLife/R/figure_5.Rmd

---
title: "Spacek et al., 2021, Figure 5"
output: pdf_document
---

```{r setup, include=FALSE}

knitr::opts_chunk$set(echo = TRUE)

library(arm)
library(lmerTest)
library(tidyverse)

source('get_data.R')
```

```{r read_data, include=FALSE}

tib = get_data("../csv/fig5.csv")
```

```{r tidy, include = FALSE}

# Transform columns st8 and opto into binary, numerical predictors
tb = tib %>% mutate(run = ifelse(st8 == "run", 1, 0))
tb = tb  %>% mutate(feedback = ifelse(opto == "TRUE", 0, 1))

tbf = tb %>% filter(feedback == 1) # feedback, panels c-f
tbs = tb %>% filter(feedback == 0) # suppression, panels i-l
```

# Figure 5c
## Effect of locomotion state on firing rate

```{r fit_model_5c}

# Random intercept, random slope for neurons,
# random intercept for experiments, nested in series
lmer.5c = lmer(rates ~ run + (1 + run | uid) + (1 | sid/eid), 
              data = tbf %>% drop_na(rates))

display(lmer.5c)
anova(lmer.5c)
```

```{r get_predicted_average_effect_5c, include=F}

mSit = fixef(lmer.5c)[1]
diffMeans = fixef(lmer.5c)[2]
mRun  = fixef(lmer.5c)[1] + diffMeans
```

Running: `r format(mRun, digits=1, nsmall=1)` spikes/s \newline
Sitting: `r format(mSit, digits=1, nsmall=1)` spikes/s \newline
n = `r nrow(tbf %>% drop_na(rates) %>% count(uid))` neurons from `r nrow(tbf %>% drop_na(rates) %>% count(mid))` mice

\newpage
# Figure 5d
## Effect of locomotion state on burst ratio

```{r fit_model_5d}

# Random intercept, random slope for neurons,
# random intercept for experiments, nested in series
lmer.5d = lmer(burstratios ~ run + (1 + run | uid) + (1 | sid/eid), 
               data = tbf %>% drop_na(burstratios))

display(lmer.5d)
anova(lmer.5d)
```

```{r get_predicted_average_effect_5d, include=F}

mSit = fixef(lmer.5d)[1]
diffMeans = fixef(lmer.5d)[2]
mRun  = fixef(lmer.5d)[1] + diffMeans
```

Running: `r format(mRun, digits=2, nsmall=2)` \newline
Sitting: `r format(mSit, digits=2, nsmall=2)` \newline
n = `r nrow(tbf %>% drop_na(burstratios) %>% count(uid))` neurons from `r nrow(tbf %>% drop_na(burstratios) %>% count(mid))` mice

\newpage
# Figure 5e
## Effect of locomotion state on sparseness

```{r tidy_for_5ef, include=FALSE}

# 'Sparseness', and 'reliability' are not computed on a trial-by-trial basis. In the csv-file, 
# these two measures are therefore identical across trials, so we simply pull out the first trial of each neuron

tbfef = tbf %>% select(mid, sid, eid, uid, mseu, run, spars, rel) %>% distinct(mseu, run, .keep_all = TRUE)

```

```{r fit_model_5e}

# Random intercept for neurons,
# random intercept for experiments, nested in series
lmer.5e = lmer(spars ~ run + (1 | uid) + (1 | sid/eid), 
               data = tbfef %>% drop_na(spars))

display(lmer.5e)
anova(lmer.5e)
```

```{r get_predicted_average_effect_5e, include=F}

mSit = fixef(lmer.5e)[1]
diffMeans = fixef(lmer.5e)[2]
mRun  = fixef(lmer.5e)[1] + diffMeans
```

Running: `r format(mRun, digits=2, nsmall=2)` \newline
Sitting: `r format(mSit, digits=2, nsmall=2)` \newline
n = `r nrow(tbfef %>% drop_na(spars) %>% count(uid))` neurons from `r nrow(tbfef %>% drop_na(spars) %>% count(mid))` mice

\newpage
# Figure 5f
## Effect of locomotion state on reliability

```{r fit_model_5f}

# Random intercept for neurons,
# random intercept for experiments, nested in series
lmer.5f = lmer(rel ~ run + (1 | uid) + (1 | sid/eid), 
               data = tbfef %>% drop_na(rel))

display(lmer.5f)
anova(lmer.5f)
```

```{r get_predicted_average_effect_5f, include=F}

mSit = fixef(lmer.5f)[1]
diffMeans = fixef(lmer.5f)[2]
mRun  = fixef(lmer.5f)[1] + diffMeans
```

Running: `r format(mRun, digits=2, nsmall=2)` \newline
Sitting: `r format(mSit, digits=2, nsmall=2)` \newline
n = `r nrow(tbfef %>% drop_na(rel) %>% count(uid))` neurons from `r nrow(tbfef %>% drop_na(rel) %>% count(mid))` mice

\newpage
# Figure 5i
## Effect of locomotion state on firing rate during V1 suppression

```{r fit_model_5i}

# Random intercept, random slope for neurons,
# random intercept for experiments nested in series
lmer.5i = lmer(rates ~ run + (1 + run | uid) + (1 | sid/eid), 
               data = tbs %>% drop_na(rates))

display(lmer.5i)
anova(lmer.5i)
```

```{r get_predicted_average_effect_5i, include=F}

mSit = fixef(lmer.5i)[1]
diffMeans = fixef(lmer.5i)[2]
mRun  = fixef(lmer.5i)[1] + diffMeans
```

Running: `r format(mRun, digits=1, nsmall=1)` spikes/s \newline
Sitting: `r format(mSit, digits=1, nsmall=1)` spikes/s \newline
n = `r nrow(tbs %>% drop_na(rates) %>% count(uid))` neurons from `r nrow(tbs %>% drop_na(rates) %>% count(mid))` mice

\newpage
# Figure 5j
## Effect of locomotion state on burst ratio during V1 suppression

```{r fit_model_5j}

# Random intercept, random slope for neurons,
# random intercept for experiments, nested in series
lmer.5j = lmer(burstratios ~ run + (1 + run | uid) + (1 | sid/eid), 
               data = tbs %>% drop_na(burstratios))

display(lmer.5j)
anova(lmer.5j)
```

```{r get_predicted_average_effect_5j, include=F}

mSit = fixef(lmer.5j)[1]
diffMeans = fixef(lmer.5j)[2]
mRun  = fixef(lmer.5j)[1] + diffMeans
```

Running: `r format(mRun, digits=2, nsmall=2)` \newline
Sitting: `r format(mSit, digits=2, nsmall=2)` \newline
n = `r nrow(tbs %>% drop_na(burstratios) %>% count(uid))` neurons from `r nrow(tbs %>% drop_na(burstratios) %>% count(mid))` mice

```{r tidy_for_5kl, include=FALSE}

# 'Sparseness', and 'reliability' are not computed on a trial-by-trial basis. In the csv-file, 
# these two measures are therefore identical across trials, so we simply pull out the first trial of each neuron

tbskl = tbs %>% select(mid, sid, eid, uid, mseu, run, spars, rel) %>% distinct(mseu, run, .keep_all = TRUE)
```

\newpage
# Figure 5k
## Effect of locomotion state on sparseness during V1 suppression

```{r fit_model_5k}

# Random intercept for neurons,
# random intercept for experiments, nested in series, nested in mice
lmer.5k = lmer(spars ~ run + (1 | uid) + (1 | mid/sid/eid), 
               data = tbskl %>% drop_na(spars))

display(lmer.5k)
anova(lmer.5k)
```

```{r get_predicted_average_change_5k, include=F}

mSit = fixef(lmer.5k)[1]
diffMeans = fixef(lmer.5k)[2]
mRun  = fixef(lmer.5k)[1] + diffMeans
```

Running: `r format(mRun, digits=2, nsmall=2)` \newline
Sitting: `r format(mSit, digits=2, nsmall=2)` \newline
n = `r nrow(tbskl %>% drop_na(spars) %>% count(uid))` neurons from `r nrow(tbskl %>% drop_na(spars) %>% count(mid))` mice

\newpage
# Figure 5l
## Effect of locomotion state on reliability during V1 suppression

```{r fit_model_5l}

# Random intercept for neurons,
# random intercept for experiments, nested in series
lmer.5l = lmer(rel ~ run + (1 | uid) + (1 | sid/eid), 
               data = tbskl %>% drop_na(rel))

display(lmer.5l)
anova(lmer.5l)
```

```{r get_predicted_average_effect_5l, include=F}

mSit = fixef(lmer.5l)[1]
diffMeans = fixef(lmer.5l)[2]
mRun  = fixef(lmer.5l)[1] + diffMeans
```

Running: `r format(mRun, digits=2, nsmall=2)` \newline
Sitting: `r format(mSit, digits=2, nsmall=2)` \newline
n = `r nrow(tbskl %>% drop_na(rel) %>% count(uid))` neurons from `r nrow(tbskl %>% drop_na(rel) %>% count(mid))` mice