---
title: "Spacek et al., 2021, Figure 5 Supplement-1"
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_5_S1ab, include=FALSE}

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

```{r tidy_for_5_S1ab, include = FALSE}

# Get relevant conditions, turn locomotion state into a binary predictor
tb = tib %>% filter(opto == "FALSE") %>% filter(st8 == "run" | st8 == "sit") %>% mutate(run = ifelse(st8 == "run", 1, 0))

# 'Signal-to-noise ratio', and 'mean peak width' 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

tbab = tb %>% select(mid, sid, eid, uid, mseu, run, snr, meanpkw) %>% distinct(mseu, run, .keep_all = TRUE)
```

# Figure 5-Supplement 1a
## Effect of locomotion state on signal-to-noise ratio (SNR)

```{r fit_model_5_S1a}

# Random intercept for neurons,
# random intercept for experiments, nested in series
lmer.5_S1a = lmer(snr ~ run + (1 | uid) + (1 | sid/eid), 
                  data = tbab %>% drop_na(snr))

display(lmer.5_S1a)
anova(lmer.5_S1a)
```

```{r predicted_average_effect_5_S1a, include=F}

mSit = fixef(lmer.5_S1a)[1]
diffMeans = fixef(lmer.5_S1a)[2]
mRun  = fixef(lmer.5_S1a)[1] + diffMeans
```

SNR locomotion: `r format(mRun, digits=2, nsmall=2)` \newline
SNR sitting: `r format(mSit, digits=2, nsmall=2)` \newline
n = `r nrow(tb %>% drop_na(snr) %>% count(uid))` neurons from `r nrow(tb %>% drop_na(snr) %>% count(mid))` mice 

\newpage
# Figure 5-Supplement 1b
## Effect of locomotion state on mean peak width

```{r fit_model_5_S1b}

# Random for neurons,
# random intercept for series
lmer.5_S1b = lmer(meanpkw ~ run + (1 | uid) + (1 | sid), 
                  data = tbab %>% drop_na(meanpkw))

display(lmer.5_S1b)
anova(lmer.5_S1b)
```

```{r predicted_average_effect_5_S1b, include=F}

mSit = fixef(lmer.5_S1b)[1]
diffMeans = fixef(lmer.5_S1b)[2]
mRun  = fixef(lmer.5_S1b)[1] + diffMeans
```

Mean peak width running: `r format(mRun, digits=2, nsmall=2)` \newline
Mean peak width sitting: `r format(mSit, digits=2, nsmall=2)` \newline
n = `r nrow(tb %>% drop_na(meanpkw) %>% count(uid))` neurons from `r nrow(tb %>% drop_na(meanpkw) %>% count(mid))` mice 

\newpage
# Figure 5-Supplement 1c
## Relation between firing rate RMI and burst ratio RMI

```{r read_data_5_S1cdefg, include=FALSE}

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

```{r tidy_for_5_S1cdefg, include=FALSE}

tb <- tib %>% filter(opto == 'FALSE')
```

```{r fit_model_5_S1c}

# Remove outliers
tb_clean <- tb %>% filter(meanburstratio < 0.99 & meanburstratio > -0.99)

# Random intercept for neurons,
# random intercept for experiments, nested in series
lmer.5_S1_c = lmer(meanburstratio ~ meanrate + (1 | uid) + (1 | sid/eid), 
                   data = tb_clean %>% drop_na(meanburstratio, meanrate))

display(lmer.5_S1_c)
anova(lmer.5_S1_c)
```

```{r save_coefficients_5_S1c, include=FALSE}

coef_df = data.frame("intercept" = fixef(lmer.5_S1_c)[1], "slope" = fixef(lmer.5_S1_c)[2], row.names = "")
write_csv(coef_df, "_stats/figure_5_S1c_coefs.csv")
```

Slope of `r format(fixef(lmer.5_S1_c)[2], digits=2, nsmall=2)` $\pm$ `r format(2 * se.fixef(lmer.5_S1_c)[2], digits=2, nsmall=2)` (95-\% confidence interval) \newline
n = `r nrow(tb_clean %>% drop_na(meanburstratio, meanrate) %>% count(uid))` neurons from `r nrow(tb_clean %>% drop_na(meanburstratio, meanrate) %>% count(mid))` mice 

\newpage
# Figure 5-Supplement 1d
## Relation between firing rate RMI and sparseness RMI

```{r fit_model_5_S1d}

# Random intercept for neurons,
# random intercept for experiments, nested in series
lmer.5_S1_d = lmer(spars ~ meanrate + (1 | uid) + (1 | sid/eid), 
                   data = tb %>% drop_na(spars, meanrate))

display(lmer.5_S1_d)
anova(lmer.5_S1_d)
```

```{r save_coefficients_5_S1d, include=FALSE}

coef_df = data.frame("intercept" = fixef(lmer.5_S1_d)[1], "slope" = fixef(lmer.5_S1_d)[2], row.names = "")
write_csv(coef_df, "_stats/figure_5_S1d_coefs.csv")
```

Slope of `r format(fixef(lmer.5_S1_d)[2], digits=2, nsmall=2)` $\pm$ `r format(2 * se.fixef(lmer.5_S1_d)[2], digits=2, nsmall=2)` (95\%-confidence interval) \newline
n = `r nrow(tb %>% drop_na(spars, meanrate) %>% count(uid))` neurons from `r nrow(tb %>% drop_na(spars, meanrate) %>% count(mid))` mice 

\newpage
# Figure 5-Supplement 1e
## Relation between firing rate RMI and reliability RMI

```{r fit_model_5_S1e}

# Random intercept for neurons,
# random intercept for experiments, nested in series,nested in mice
lmer.5_S1e = lmer(rel ~ meanrate + (1 | uid) + (1 | mid/sid/eid), 
                  data = tb %>% drop_na(rel, meanrate))

display(lmer.5_S1e)
anova(lmer.5_S1e)
```

```{r store_coefficients_5_S1e, include=F}

coef_df = data.frame("intercept" = fixef(lmer.5_S1e)[1], "slope" = fixef(lmer.5_S1e)[2], row.names = "")
write_csv(coef_df, "_stats/figure_5_S1e_coefs.csv")
```

Slope of `r format(fixef(lmer.5_S1e)[2], digits=2, nsmall=2)` $\pm$ `r format(2 * se.fixef(lmer.5_S1e)[2], digits=2, nsmall=2)` (95\%-confidence interval) \newline
n = `r nrow(tb %>% drop_na(rel, meanrate) %>% count(uid))` neurons from `r nrow(tb %>% drop_na(rel, meanrate) %>% count(mid))` mice 

\newpage
# Figure 5-Supplement 1f
## Relation between firing rate RMI and SNR RMI

```{r fit_model_5_S1f}

# Random intercept for neurons,
# random intercept for experiments, nested in series
lmer.5_S1f = lmer(snr ~ meanrate + (1 | uid) + (1 | sid/eid), 
                  data = tb %>% drop_na(snr, meanrate))

display(lmer.5_S1f)
anova(lmer.5_S1f)
```

```{r store_coefficents_5_S1f, include=F}

coef_df = data.frame("intercept" = fixef(lmer.5_S1f)[1], "slope" = fixef(lmer.5_S1f)[2], row.names = "")
write_csv(coef_df, "_stats/figure_5_S1f_coefs.csv")
```

Slope of `r format(fixef(lmer.5_S1f)[2], digits=2, nsmall=2)` $\pm$ `r format(2 * se.fixef(lmer.5_S1f)[2], digits=2, nsmall=2)` (95\%-confidence interval) \newline
n = `r nrow(tb %>% drop_na(snr, meanrate) %>% count(uid))` neurons from `r nrow(tb %>% drop_na(snr, meanrate) %>% count(mid))` mice 

\newpage
# Figure 5-Supplement 1g
## Relation between firing rate RMI and peak width RMI

```{r fit_model_5_S1g}

# Random intercept for mice
lmer.5_S1g = lmer(meanpkw ~ meanrate + (1 | mid), 
                  data = tb %>% drop_na(meanpkw, meanrate))

display(lmer.5_S1g)
anova(lmer.5_S1g)
```

```{r store_coefficients_5_S1g, include=FALSE}

coef_df = data.frame("intercept" = fixef(lmer.5_S1g)[1], "slope" = fixef(lmer.5_S1g)[2], row.names = "")
write_csv(coef_df, "_stats/figure_5_S1g_coefs.csv")
```

Slope of `r format(fixef(lmer.5_S1g)[2], digits=2, nsmall=2)` $\pm$ `r format(2 * se.fixef(lmer.5_S1g)[2], digits=2, nsmall=2)` (95\%-confidence interval) \newline
n = `r nrow(tb %>% drop_na(meanpkw, meanrate) %>% count(uid))` neurons from `r nrow(tb %>% drop_na(meanpkw, meanrate) %>% count(mid))` mice 

\newpage
# Figure 5-Supplement 1h
# Distributions of eye position variability, separated by locomotion state

```{r read_data_5_S1h, include=FALSE}

tib = get_data("../csv/ipos_st8.csv")

# Turn locomotion state into a binary predictor
tb = tib %>% mutate(run = ifelse(st8 == "run", 1, 0))

# 'Standard deviation of eye position is not computed on a trial-by-trial basis. In the csv-file, 
# this measure is therefore identical across trials, so we simply pull out the first trial of each experiment

tbh = tb %>% select(mid, sid, eid, mse, run, std_xpos_cross) %>% distinct(mse, run, .keep_all = TRUE)

```

```{r fit_model_5_S1h}

# Random intercept for series
lmer.5_S1h = lmer(std_xpos_cross ~ run + (1 | sid), 
                  data = tbh %>% drop_na(std_xpos_cross))

display(lmer.5_S1h)
anova(lmer.5_S1h)
```

```{r predicted_average_effect_5_S1h, include=F}

mSit = fixef(lmer.5_S1h)[1]
diffMeans = fixef(lmer.5_S1h)[2]
mRun  = fixef(lmer.5_S1h)[1] + diffMeans
```

Locomotion: mean eye position standard deviation of `r format(mRun, digits=2, nsmall=2)` (95\%-confidence interval) \newline
Sitting:  mean eye position standard deviation of `r format(mSit, digits=2, nsmall=2)` \newline
n = `r nrow(tbh %>% drop_na(std_xpos_cross) %>% count(eid))` experiments from `r nrow(tbh %>% drop_na(std_xpos_cross) %>% count(mid))` mice  

\newpage
# Figure 5-Supplement 1i
# Relation between locomotion effects on eye position variability and firing rate variability

```{r read_data_5_S1i, include=FALSE}

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

```{r fit_model_5_S1i}

# Random effect of experiment, units partially crossed
lmer.5_S1i = lmer(relrmi ~ iposrmi + (1 | uid), 
                  data = tib %>% drop_na(relrmi, iposrmi))

display(lmer.5_S1i)
anova(lmer.5_S1i)
```

```{r save_coefficients_5_S1i, include=FALSE}

coef_df = data.frame("intercept" = fixef(lmer.5_S1i)[1], "slope" = fixef(lmer.5_S1i)[2], row.names = "")
write_csv(coef_df, "_stats/figure_5_S1i_coefs.csv")

```


Slope of `r format(fixef(lmer.5_S1i)[2], digits=2, nsmall=2)` $\pm$ `r format(2 * se.fixef(lmer.5_S1i)[2], digits=2, nsmall=2)` (95\%-confidence interval)

```{r assess_coefficient, include = FALSE}

foo = tib %>% filter(iposrmi != "NA")
sd_ipos = sd(foo$iposrmi)

# expected difference in rel RMI corresponding to a 1-sd difference in eye pos signm
exp_diff = fixef(lmer.5_S1i)[2] * sd_ipos

print(paste("The expected difference in relRMI corresponding to a 1 standard deviation difference in eyePosSigmaRMI is ", format(exp_diff, digits=2)))

```
Expected difference in reliability RMI corresponding to a 1 standard deviation difference in eye position $\sigma$ RMI is `r format(exp_diff, digits=2, nsmall=2)`, the standard deviation of the residuals is `r format(sigma(lmer.5_S1i), digits=2, nsmall=2)`. \newline \newline
n = `r nrow(tib %>% drop_na(relrmi, iposrmi) %>% count(uid))` neurons from `r nrow(tib %>% drop_na(relrmi, iposrmi) %>% count(mid))` mice