Overview

Trio (or familial) analysis has been exceptionally powerful for identifying rare childhood diseases.  The most prominent publication in this area is this first example of whole exome sequencing saving a life.  There are many other publications since and some review articles such as this one.  Familial analysis is critical for rare, autosomal dominant diseases because, almost by definition, the mutations may be "private" to each individual so we can't look across big populations to find one single causative mutation.  But within families, we can identify bad private mutations in single genes or pathways and then look across populations to find commonality at the gene or pathway level to explain a phenotype.

Learning Objectives

1. Review differences in calling mutations on higher level organisms
2. Call SNVs from multiple samples simultaneously
3. Determine relationship between the individuals based on the results

Diploid genomes

The initial steps in calling variants for diploid or multi-ploid organisms with NGS data are the same as what we've already seen:

1. Map the raw data to a suitable reference. 
2. Look for SNVs.

Expectations of output are quite different however, which can add statistical power to uncovering variation in populations or organisms with more than two expected variants at the same location. If you happen to be working with a model organism with extensive external data (ESPECIALLY HUMAN), then there are even more sophisticated tools like the Broad Institute's GATK that can improve both sensitivity and specificity of your variant calls.

Get some data

Many example datasets are available from the 1000 Genomes Project specifically for method evaluation and training. We'll explore a trio (mom, dad, child). Known as the CEU Trio from the 1000 genomes project  their accession numbers are NA12892, NA12891, and NA12878 respectively. To make the exercise run FAR more quickly, we'll focus on data only from chromosome 20.

All the data we'll use is located here:

mkdir $SCRATCH/GVA_Human_trios
cd $SCRATCH/GVA_Human_trios
cp -r $BI/ngs_course/human_variation raw_files
ls raw_files

This directory contains the following:

1. compressed raw data (the .fastq.gz files)
2. mapped data (the .bam files)
3. variant calls (the .vcf files)
4. the subdirectory ref with special references
5. .bam files containing a subset of mapped human whole exome data are also available on these three; those are the three files "NA*.bam".
6. We've pre-run samtools and GATK on each sample individually - those are the *GATK.vcf and *samtools.vcf files.
7. We've also pre-run samtools and GATK on the trio, resulting in GATK.all.vcf and samtools.all.vcf. (these files are from old versions)
8.  The 1000 Genomes project is really oriented to producing .vcf files; the file "ceu20.vcf" contains all the latest genotypes from this trio based on abundant data from the project. 

Single-sample variant calling with bcftools

We would normally use the BAM file from a previous mapping step to call variants in this raw data. However, for the purposes of this course we will use the actual BAM file provided by the 1000 Genomes Project (from which the .fastq file above was derived, leading to some oddities in it). As a bonus tutorial, you could map the data yourself and using what you learned in the bowtie2 tutorial and then use the resultant .bam files.

For now, the bam file we want to focus on is:

$SCRATCH/GVA_Human_trios/raw_files/NA12878.chrom20.ILLUMINA.bwa.CEU.exome.20111114.bam

With bcftools, this is a two-step process:

1. bcftools mpileup command transposes the mapped data in a sorted BAM file fully to genome-centric coordinates. It starts at the first base on the first chromosome for which there is coverage and prints out one line per base. Each line has information on every base observed in the raw data at that base position along with a lot of auxiliary information depending on which flags are set. It calculates the Bayseian prior probability given by the data, but does not estimate a real genotype.
2. bcftools call with a few options added uses the prior probability distribution and the data to calculate a genotype for the variants detected.

Recall that we installed samtools and bcftools in our GVA-SNV conda environment. Make sure you have activated your GVA-SNV environment and you have access to samtools and bcftools version 1.15.1

conda activate GVA-SNV

samtools --version
bcftools --version

samtools 1.15.1
Using htslib 1.15.1
Copyright (C) 2022 Genome Research Ltd.
# Followed by a bunch of compilation details.

bcftools 1.15.1
Using htslib 1.15.1
Copyright (C) 2022 Genome Research Ltd.
License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl.html>
This is free software: you are free to change and redistribute it.
There is NO WARRANTY, to the extent permitted by law.

cd $SCRATCH/GVA_Human_trios 
bcftools mpileup --threads 64 -O u -f raw_files/ref/hs37d5.fa raw_files/NA12878.chrom20.ILLUMINA.bwa.CEU.exome.20111114.bam | bcftools call --threads 64 -v -c - > trios_tutorial.raw.vcf

The above command may take ~20 minutes to run based on some student's experience and will not produce many lines of output leading students to worry their terminal has locked up. More than likely this is not the case, but you can try hitting the return key 1 or 2 times to see if your terminal adds a blank line to verify. As this command is taking so long it is recommended that you read ahead and/or switch to reading another tutorial while waiting for this command to finish.

One potential issue with this type of approach is that vcf files only record variation that can be seen with the data provided. When all reads mapping to a given location exactly match the reference (i.e. is homozygous wildtype relative to the reference) there will be no data. Which looks the same as if you had no data in those regions; this leads us to our next topic. 

Multiple-sample variant calling with bcftools

Not being able to tell between no data and wildtype is not the end of the world for a single sample, but if you're actually trying to study human (or other organism) genetics, you must discriminate homozygous WT from a lack of data. This is done by providing many samples to the variant caller simultaneously. This concept extends further to populations; calling variants across a large and diverse population provides a stronger Bayesian prior probability distribution for more sensitive detection.

To instruct bcftools to call variants across many samples, you must simply give it mapped data with each sample tagged separately. bcftools allows two methods to do this:

1. By providing separate bam files for each sample. Note that in the following code block, he trailing \ symbols tells the command line that you are not done entering the command and not to execute any commands until you hit return without a preceding \ mark.

   Warning, DO NOT run this command yet.

   bcftools multi-sample variants: separate bam files

   #bcftools mpileup --threads 64 -O u -f raw_files/ref/hs37d5.fa \
     raw_files/NA12878.chrom20.ILLUMINA.bwa.CEU.exome.20111114.bam \
     raw_files/NA12891.chrom20.ILLUMINA.bwa.CEU.exome.20111114.bam \
     raw_files/NA12892.chrom20.ILLUMINA.bwa.CEU.exome.20111114.bam \
       | bcftools call -v -c - > trios_tutorial.multi-sample.vcf
   

   
   The output file from this option is $BI/gva_course/GVA.multi-sample.vcf if you want to work with it without having to wait on the analysis to run personally.

   
   

2. By providing one or more bam files, each containing mapped reads from multiple samples tagged with unique samtools @RG tags.

   Do not run this command

   This command comes with a sizable caveat: If you intend to use this option, you must make sure you tag your reads with the right RG tag; this can easily be done during the samse or sampe stage of mapping with bwa with the -r option, using samtools merge, with picard tools AddOrReplaceReadGroup command, or with your own perl/python/bash commands. The problem is that it requires prior knowledge of all the samples you are going to group together ahead of time, and individual sample input files have to be merged together. There are obviously times that this will be the desired and easier, but for the purpose of this course, we believe it makes more sense to keep all input files separate and only have a single merged output file. As part of the tutorial we do provide the appropriate files necessary for the following command to run.

   samtools multi-sample variants: one or more bam files using @RG

   bcftools mpileup --threads 64 -O u -f raw_files/ref/hs37d5.fa <all_with@RG.bam> | bcftools call --threads 68 -v -c - > trios_tutorial.all.raw.vcf 

   An observation for the individuals interested in phylogenies

   I have a hunch that the use of read tags could be useful for your analysis IF you have a finite set of samples that you intend to analyze, have all the data available at once, and are willing to repeat your analysis from near scratch in the event of either of these factors changing. This is an excellent example of the importance of finding papers that have done similar analyses, and mimicking as much of what they have done as possible/reasonable.

Based on the discussion above, we are selecting the first solution and providing details of how this command should be run. As this command will generate very little output and take ~30 minutes to complete, you are once again reminded that the output file from this option is available $BI/gva_course/GVA.multi-sample.vcf if you want to work with it without having to wait on the analysis to run personally.

cd $SCRATCH/GVA_Human_trios
bcftools mpileup --threads 64 -O u -f raw_files/ref/hs37d5.fa \
  raw_files/NA12878.chrom20.ILLUMINA.bwa.CEU.exome.20111114.bam \
  raw_files/NA12891.chrom20.ILLUMINA.bwa.CEU.exome.20111114.bam \
  raw_files/NA12892.chrom20.ILLUMINA.bwa.CEU.exome.20111114.bam \
    | bcftools call -v -c - > trios_tutorial.multi-sample.vcf 

The observant students who start this command might notice that this time mpileup does give a message that it is working with 3 samples, and 3 input files (while all previous analysis have used 1 file and 1 sample).

Identify the lineage

If genetics works, you should be able to identify the child based strictly on the genotypes.  Can you do it?

cat trios_tutorial.multi-sample.vcf | tail -10000 | awk '{if ($6>500) {print $2"\t"$10"\t"$11"\t"$12}}' | grep "0/0" | sed s/':'/'\t'/g | awk '{print $2"\t"$4"\t"$6}' | tail -100 | sort | uniq -c | sort -n -r

     35 0/1	0/1	0/0
     20 0/0	0/1	0/0
     18 0/1	0/0	0/1
     14 0/1	1/1	0/0
      6 0/0	0/1	0/1
      5 1/1	1/1	0/0
      1 1/1	0/1	0/0
      1 0/1	0/0	0/0

Given this information can you make any determination about family structure for these 3 individuals? The first column is the number of occurrences generated by the uniq -c command in our large 1 liner, with the following 3 columns being the different individual samples. So consider that while some lines may not show a viable mendelian inheritance pattern, you should weight things according to how many times each scenario occurred as our filtering was fairly limited.

     35 0/1	0/1	0/0
     20 0/0 0/1 0/0
     18 0/1	0/0	0/1
     14 0/1	1/1	0/0	# middle can't be child
      6 0/0	0/1	0/1
      5 1/1	1/1	0/0	* No mendelian combination exists
      1 1/1	0/1	0/0	*
      1 0/1	0/0	0/0	*

Going further

Refining your analysis

Can you modify the large 1 liner command to be more strict, or to include more examples such that you eliminate the non-mendelian inheritance situations and consider a larger number of loci?

Scope and usefulness

• This same type of analysis can be done on much larger cohorts, say cohorts of 100 or even 1000s of individuals with known disease state to attempt to identify associations between allelic state and disease state.
• This is the first step of building a phylogeny of 3 related individuals. Expanding this could be of use for genetic counseling or larger phylogenetic analysis.

Return to GVA2022 page.