MicrobeCensus 1.0.4

Estimation of average genome size from metagenomic data

# MicrobeCensus
MicrobeCensus is a fast and easy to use pipeline for estimating the average genome size (AGS) of a microbial
community from metagenomic data.

In short, AGS is estimated by aligning reads to a set of universal single-copy gene families present in nearly all cellular microbes (Bacteria, Archaea, Fungi).
Because these genes occur once per genome, the average genome size of a microbial community is inversely proportional to the fraction of reads which hit these genes.

Once AGS is obtained, it becomes possible to obtain the total coverage of microbial genomes present in a sample (genome equivalents = total bp sequenced/AGS in bp), which can be useful for normalizing gene abundances.

### Requirements
* Python dependencies (installed via setup.py): Numpy, BioPython
* Supported platforms: Mac OSX, Unix/Linux; Windows not currently supported
* Python version 2 or 3

### Installation
Download MicrobeCensus from: https://github.com/snayfach/MicrobeCensus/archive/v1.0.4.tar.gz

Unpack the project:
`tar -zxvf MicrobeCensus-1.0.4.tar.gz`

Navigate to the installation directory:
`cd /path/to/MicrobeCensus`

Run setup.py. This will install any dependencies:
`python setup.py install` or
`sudo python setup.py install` to install as a superuser

Alternatively, MicrobeCensus can be installed directly from PyPI:
`pip install MicrobeCensus` or
`sudo pip install MicrobeCensus` to install as a superuser

### Using MicrobeCensus without installing
Although this is not recommended, users may wish to run MicrobeCensus without running setup.py.

Both BioPython and Numpy will both need to be already installed.
You should be able to enter the following command in the python interpreter without getting an error:
`>>> import Bio.SeqIO`
`>>> import numpy`

Next, add the MicrobeCensus module to your PYTHONPATH environmental variable:
`export PYTHONPATH=$PYTHONPATH:/path/to/MicrobeCensus` or
`echo -e "\nexport PYTHONPATH=\$PYTHONPATH:/path/to/MicrobeCensus" >> ~/.bash_profile` to avoid entering the command in the future

Finally, add the scripts directory to your PATH environmental variable:
`export PATH=$PATH:/path/to/MicrobeCensus/scripts` or
`echo -e "\nexport PATH=\$PATH:/path/to/MicrobeCensus/scripts" >> ~/.bash_profile` to avoid entering the command in the future

Now, you should be able to enter the command into your terminal without getting an error:
`run_microbe_census.py -h`

### Testing the software
After installing MicrobeCensus, we recommend testing the software:
`cd /path/to/MicrobeCensus/test`
`python test_microbe_census.py`

### Running MicrobeCensus
MicrobeCensus can either be run as a command-line script or imported to python as a module.

#### Command-line usage
**run_microbe_census.py [-options] seqfile outfile**

Input/Output (required):
* **seqfile:** path to input metagenome; can be FASTQ/FASTA; can be
gzip (.gz) or bzip (.bz2) compressed
* **outfile:** path to output file containing AGS estimate

Pipeline throughput (optional):
* **-n NREADS:** number of reads to sample from seqfile and use for AGS
estimation. to use all reads set to 100000000.
(default = 1e6)
* **-t THREADS:** number of threads to use for database search (default= 1)

File type (optional):
* **-f {fasta,fastq}:** file type (default = autodetect)
* **-c {fastq-sanger,fastq-solexa,fastq-illumina}:** quality score encoding (default = autodetect)

Quality control (optional):
* **-l {50,60,70,80,90,100,110,120,130,140,150,175,200,225,250,300,350,400,450,500}:**
all reads trimmed to this length; reads shorter than
this discarded (default = median read length)
* **-q MIN_QUALITY:** minimum base-level PHRED quality score (default = -5;
no filtering)
* **-m MEAN_QUALITY:** minimum read-level PHRED quality score (default = -5;
no filtering)
* **-d:** filter duplicate reads (default = False)
* **-u MAX_UNKNOWN:** max percent of unknown bases perread (default = 100
percent; no filtering)

Misc options:
* **-h, --help:** show this help message and exit
* **-v:** print program's progress to stdout (default = False)
* **-V, --version:** show program's version number and exit

#### Module usage

First, import the module:
`>>> from microbe_census import microbe_census`

Next, setup your options and arguments, formatted as a dictionary. The path to your metagenome is the only requirement (default values will be used for all other options):
`>>> args = {'seqfile':'MicrobeCensus/microbe_census/example/example.fq.gz'}`

Alternatively, other options can be specified:
```
>>> args = {
'seqfile':'MicrobeCensus/microbe_census/example/example.fq.gz',
'nreads':100000,
'read_length':100,
'file_type':'fastq',
'quality_type':'fastq-sanger',
'threads':1,
'min_quality':10,
'mean_quality':10,
'filter_dups':False,
'max_unknown':0,
'verbose':True}
```

Finally, the entire pipeline can be run by passing your arguments to the run_pipeline function. MicrobeCensus returns the estimated AGS of your metagenome, along with a dictionary of used arguments:
`average_genome_size, args = microbe_census.run_pipeline(args)`

#### Recommended options
* When in doubt, use default parameters! In most cases, MicrobeCensus tries to pick the best parameters for you.
* For more accurate estimates of AGS, use -n to increase the number of reads sampled. The default value of 1,000,000 should give good results, but more reads may result in slightly more accurate estimates, particularly when AGS is very large.
* Don't use quality filtering options (-q, -m, -d, -u) if you plan on using MicrobeCensus for normalization. In this case, MicrobeCensus should be directly run on the metagenome you used for estimating gene-family abundances.
* Use -v/--verbose to print program progress

### Output format
The output is a tab delimited field with a header line and three fields:
* **reads_sampled**: this is the number of reads sampled from the metagenome to estimate AGS. it is likely that this is less than the actual number of reads in the metagenome
* **read_length**: all reads were trimmed to this length. reads shorter than this were discarded. this is different from the actual read length of your metagenome
* **avg_size**: the average genome size (in bp) of your input metagenome

### Normalization
Once AGS is obtained, it becomes trivial to obtain the total coverage of microbial genomes present in a metagenome.

>genome equivalents = (total DNA sequenced in bp)/(average genome size in bp), and
>total DNA sequenced in bp = (read length in bp) * (reads sequenced)

**_Note: you will need to compute the total DNA sequenced in bp on your own! This value is not provided by MicrobeCensus and cannot be obtained using the MicrobeCensus output._**

The number of genome equivalents can then be used to normalize count data obtained from metagenomes using the statistic **RPKG (reads per kb per genome equivalent)**. This is similar to the commonly used statistic RPKM, but instead of dividing by the number of total mapped reads, we divide by the number of genome equivalents:

>RPKG = (reads mapped to gene)/(gene length in kb)/(genome equivalents)

Use case:
We have two metagenomic libraries, L1 and L2, which each contain 1 million 100-bp reads:
>READ_LENGTH_L1 = 100 bp
>READS_SEQUENCED_L1 = 1,000,000
>TOTAL_DNA_L1 = 100,000,000 bp
>READ_LENGTH_L2 = 100 bp
>READS_SEQUENCED_L2 = 1,000,000
>TOTAL_DNA_L2 = 100,000,000 bp

We use MicrobeCensus to estimate the average genome size of each library:
>AGS_L1 = 2,500,000 bp
>AGS_L2 = 5,000,000 bp

Next, we map reads from each library to a reference database which contains a gene of interest G. G is 1000 bp long.
We get 100 reads mapped to gene G from each library:
>LENGTH_G = 1,000 bp
>MAPPED_READS_G_L1 = 100
>MAPPED_READS_G_L2 = 100

Finally, we quantify RPKG for gene G in each library:
>RPKG for G in L1 = (100 mapped reads)/(1 kb)/(100,000,000 bp sequenced / 2,500,000 bp AGS) = 2.5
>RPKG for G in L2 = (100 mapped reads)/(1 kb)/(100,000,000 bp sequenced / 5,000,000 bp AGS) = 5.0

### Software speed
* Run times are for a 150 bp library. Expect longer/shorter runtimes depending on read length.

Threads (-t) | Reads/Second
------------- | -------------
1 | 830
2 | 1,300
4 | 1,800
8 | 2,000

### Training
We have included scripts and documentation for retraining MicrobeCensus, using user-supplied training genomes and gene families. Documentation and scripts can be found under: MicrobeCensus/training

### Citing
If you use MicrobeCensus, please cite:

Nayfach, S. and Pollard, K.S. Average genome size estimation improves comparative metagenomics and sheds light on the functional ecology of the human microbiome. _Genome biology 2015;**16**(1):51_.