# PSMC 

Pairwise Sequentially Markovian Coalescent (PSMC) model is a method used in genetics to study the history of populations. It analyses the genetic variation within the DNA of a single individual to infer past population sizes over time. It is a way to look back in time and see how the number of ancestors of a species has changed, which can give insights into events like population bottlenecks or expansions.

Here we use the software package from https://github.com/lh3/psmc

## What you need
PSMC requires **a diploid FASTQ (.fq) file** as input, this is generated as part of the SwarmGenomics pipeline in [03.Selecting largest chromosomes](https://github.com/AureKylmanen/Swarmgenomics/blob/main/03.%20Selecting%20largest%20chromosomes.md). Alternatively, you can either use a FASTQ you already have or generate one from a VCF file according to the instructions.

### Installations

#### PSMC

In your software directory (e.g. /vol/storage/software)
```
#clone the repository
git clone https://github.com/lh3/psmc.git

#make psmc
cd psmc
make

#make utils
cd utils
make
```
## Running PSMC.bash
Download [**psmc.bash**](https://github.com/AureKylmanen/Swarmgenomics/blob/main/Scripts/psmc.bash) and [**params.txt**](https://github.com/AureKylmanen/Swarmgenomics/blob/main/Parameters/params.txt).

Edit the params.txt if you wish to change the PSMC parameters. Note that a generation time of 25 years might be appropriate for humans, but many organisms have a smaller generation time.
```
# ============================
# PSMC
# ============================
DIPLOID_FASTQ="${WORKING_DIR}/${SPECIES}/diploid.fq.gz"
DIPLOID_PSMCFA="${WORKING_DIR}/${SPECIES}/diploid.psmcfa"
DIPLOID_PSMC="${WORKING_DIR}/${SPECIES}/diploid.psmc"

PSMC_N_ITER=25
PSMC_T=15
PSMC_R=5
PSMC_PATTERN="2+2+25*2+4+6"

MUTATION_RATE=1.25e-8
GENERATION_TIME=25
```
Remember to ```chmod +x psmc.bash```. 
Then run with:
```
./psmc.bash "species" "params.txt"
```
The results will be copied into the results directory within the species directory.

## Running PSMC step-by-step

#### Convert fastq files to PSMC fasta format
```
/path/to/psmc/utils/fq2psmcfa -q20 /working_dir/diploid.fq.gz > /working_dir/diploid.psmcfa

# Example
# /vol/storage/software/psmc/utils/fq2psmcfa -q20 /vol/storage/swarmGenomics/golden_eagle/diploid.fq.gz > /vol/storage/swarmGenomics/golden_eagle/diploid.psmcfa
```
#### Run the PSMC analysis
You may change the parameters according to your preferences. This step will take a few hours, so you may wish to use "nohup" and "&".
```
nohup psmc -N25 -t15 -r5 -p "2+2+25*2+4+6" -o diploid.psmc diploid.psmcfa &

# Example
# nohup /vol/storage/software/psmc/psmc -N25 -t15 -r5 -p "2+2+25*2+4+6" -o /vol/storage/swarmGenomics/golden_eagle/diploid.psmc /vol/storage/swarmGenomics/golden_eagle/diploid.psmcfa &
```
#### Convert the output of PSMC analysis
```
utils/psmc2history.pl diploid.psmc | utils/history2ms.pl > ms-cmd.sh

# Example
# /vol/storage/software/psmc/utils/psmc2history.pl /vol/storage/swarmGenomics/golden_eagle/diploid.psmc | /vol/storage/psmc_plot_dir/psmc/utils/history2ms.pl > /vol/storage/swarmGenomics/golden_eagle/ms-cmd.sh
```

### Plot the results
```
# Plot as pdf with default parameters
/vol/storage/software/psmc/utils/psmc_plot.pl diploid diploid.psmc

# Example
# /vol/storage/psmc_plot_dir/psmc/utils/psmc_plot.pl -p /vol/storage/swarmGenomics/golden_eagle/diploid diploid.psmc


# OR if you know the generation time and mutation rate, you can plot with the specific values

/vol/storage/software/psmc/utils/psmc_plot.pl -u <mutation_rate> -g <generation_time> diploid diploid.psmc

# Example:
# utils/psmc_plot.pl -u 1.29e-8 -g 12 diploid diploid.psmc

```
## PSMC output
PSMC estimates historical effective population size (Ne) from a diploid genome. The main output is a .psmc file (numeric estimates) and a plot showing Ne over time.
- X-axis: Time in generations (can convert to years using generation time and mutation rate).

- Y-axis: Effective population size (Ne).

Interpretation: Peaks = population expansions, troughs = bottlenecks.

**Note:** PSMC captures long-term trends and may be less accurate for very recent events.