asTair is a toolchain to process DNA modification sequencing data. asTair was designed primarily for handling TET-Assisted Pyridine Borane (TAPS) sequencing output, but also contains functions that are useful for Bisulfite Sequencing (BS) data.

Installation through pip is the easiest way to get asTair, and it works in python2 and 3:

pip install astair

You should now be able to call astair:

astair --help

Usage: astair [OPTIONS] COMMAND [ARGS]...

  asTair (tools for processing cytosine modification sequencing data)

Options:
  --help  Show this message and exit.

Commands:
  align     Align raw reads in fastq format to a reference genome.
  call      Call modified cytosines from a bam or cram file.
  filter    Look for sequencing reads with more than N CpH modifications.
  find      Output positions of Cs from fasta file per context.
  idbias    Generate indel count per read length information (IDbias).
  mbias     Generate modification per read length information (Mbias).
  phred     Calculate per base (A, C, T, G) Phred scores for each strand.
  simulate  Simulate TAPS/BS conversion on top of an existing bam/cram file.
  summarise  Collects and outputs modification information per read.

  __________________________________About__________________________________
  asTair was written by Gergana V. Velikova and Benjamin Schuster-Boeckler.
  This code is made available under the GNU General Public License, see
  LICENSE.txt for more details.
                                                           Version: 3.x.x

In general, you can use --help on all astair sub-commands to get detailed instructions on the available options.

(If for some reason pip is not an option, check our FAQ for further ways to install asTair.)

All of the examples in the main part of the current tutorial are based on the assumption that the input sequencing data are TAPS pair-end sequencing reads, however, asTair analyses can be run in single-end mode (--se). Also, asTair enables you to run analyses on WGBS data, which requires a running installation of bwa-meth for the alignment step. For more information on WGBS analyses you may check the section Analysis of WGBS data (or other unmodified cytosine to thymine conversion methods).

We will assume that you have generated paired-end sequencing data, which is stored in two fastq files. For this brief tutorial, we assume the files are called lambda.phage_test_sample_R1.fq.gz and lambda.phage_test_sample_R2.fq.gz. If you want to follow this tutorial, you can download the files here:

# Or use curl -O if wget is not available
wget https://zenodo.org/record/2582855/files/lambda.phage_test_sample_1.fq.gz
wget https://zenodo.org/record/2582855/files/lambda.phage_test_sample_2.fq.gz

The raw reads need to be aligned. asTair contains a command to help with this. It assumes that bwa and samtools are available on your system. (If you prefer to use a different aligner, skip to step 2.)

You will also need an indexed reference genome to align to, which can be given as a bgzip compressed file. For this example we are using the lambda phage genome, which you can download with

wget https://zenodo.org/record/2582855/files/lambda_phage.fa
wget https://zenodo.org/record/2582855/files/lambda_phage.fa.fai

Now, you are ready to align:

mkdir -p output_dir
astair align -f lambda_phage.fa -1 lambda.phage_test_sample_1.fq.gz -2 lambda.phage_test_sample_2.fq.gz -d output_dir

If the reference FASTA file contains spaces in the header, algnment and calling will proceed using only the first word in the description unless the parameters '--add_undescores' and '--use_underscores' (aligner only) are used.

Once your fastq files are aligned and sorted (done automatically by astair align), you can run astair call to create a list of putative modified positions:

astair call -i output_dir/lambda.phage_test_sample_mCtoT.cram -f lambda_phage.fa --context CpG --minimum_base_quality 13 -d output_dir/

You can skip positions from the 5' or 3' of the reads if they seem to show Mbias by the --start_clip and --end_clip options. NB: In case positions are poorly covered or are covered only by reads' start and end positions, the usage of --start_clip and --end_clip can can alter the modified/unmodified ratio or show the genomic position as uncovered. The --no_information option is also critical and we recommend to use 0, followed by a filtering of positions by the sum of modified and unmodified positions to be greater than 0.

After calling methylation, you will find two additional files in output_dir:

1. lambda.phage_test_sample_mCtoT_mCtoT_CpG.stats
2. lambda.phage_test_sample_mCtoT_mCtoT_CpG.mods

The .stats file contains global statistics on the modification rate in different sequence contexts. You can use this to get an idea of the overall level of modification in your sample. Here you will find information about how many cytosine positions of certain context are in the reference, how many of them were covered, and how many reads were modified/unmodified at the covered positions on the relevant strand assuming directionality. In our example here, we used a 1:1 mixture of in-vitro modified and unmodified lambda phage, so the results show a methylation rate of approximately 50% :

The .mods file contains per-position information on your sample:

The header should be mostly self-explanatory. MOD and UNMOD refer to the number of reads covering that base that showed evidence of modification/no modification, and were of the right orientation to be meaningful for modification calling. The total coverage, including reads that were oriented in a way that no modification information can be extracted, is shown in TOTAL_DEPTH. SNV gives a heuristic indication whether the position is indeed a modified base, or a genetic C to T variant in the genome of the sample.

1. Do quality control of the sequencing reads and do quality trimming before mapping and dispose of very short reads, using FastQC, trimgalore or similar tools.
2. In most cases, it will be best to remove PCR duplicates before running the modification caller, unless your reads are non-randomly fragmented (e.g. enzymatically digested).
3. Check the fragment (insert) size distribution and decide on an overlap removal method for paired-end reads. The simplest option is the default removal of overlaps handled by astair call, which will randomly select one of two overlapping reads. This behaviour can be disabled by the -sc option, in case you are using a more sophisticated overlap-clipping tool.
4. For speed and convenience we recommend using the --per_chromosome option, if possible, in order to run multiple processes in parallel. This also reduces the memory requirement when asTair is run on a desktop machine.

The analysis pipeline for bisulfite sequencing data does follows the same steps as TAPS data analysis, but requires different options. We again start from fastq files. To avoid Bismark-style double-alignments, we prefer to use bwa meth, which can be used directly through astair align when you choose the --method CtoT option.

mkdir -p output_dir
astair align -f lambda_phage.fa -1 lambda.phage_test_sample_BS_1.fastq.gz -2 lambda.phage_test_sample_BS_2.fastq.gz --method CtoT -d output_dir/

You can now use astair call with --method CtoT for the modifcation calling:

astair call -i output_dir/lambda.phage_test_sample_BS_CtoT.cram -f lambda_phage.fa --method CtoT --context CpG --minimum_base_quality 13 -d output_dir/

• More information on other asTair tools
• asTair FAQ

This software is made available under the terms of the GNU General Public License v3.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, TITLE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE COPYRIGHT HOLDERS OR ANYONE DISTRIBUTING THE SOFTWARE BE LIABLE FOR ANY DAMAGES OR OTHER LIABILITY, WHETHER IN CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.