iSAFE: integrated Selection of Allele Favored by Evolution

Most approaches that capture signatures of selective sweeps in population genomics data do not identify the specific mutation favored by selection. The iSAFE enables researchers to accurately pinpoint the favored mutation in a large region (∼5 Mbp) by using a statistic derived solely from population genetics signals. iSAFE does not require knowledge of demography, the phenotype under selection, or functional annotations of mutations.

• Akbari, Ali, Joseph J. Vitti, Arya Iranmehr, Mehrdad Bakhtiari, Pardis C. Sabeti, Siavash Mirarab, and Vineet Bafna. "Identifying the favored mutation in a positive selective sweep." Nature Methods (2018).
• Contact: Ali Akbari (Ali_Akbari@hms.harvard.edu). If I can be of assistance, please do not hesitate to contact me.

FAQ – frequently asked questions

Please read the FAQ for answers to the most common queries.

Software requirements

1. Python2.7 and the following packages are required:
   • numpy version 1.9 or above
   • pandas version 0.18 or above
2. bcftools version 1.2 or above (only for --format vcf, not required if you are using --format hap).
   • Follow the bcftools installation guideline.
   • iSAFE assumes the bcftools binary file is installed to a bin subdirectory that is added to your $PATH; otherwise, you have to change the following line in ./src/bcftools.py to the bcftools binary file path:

   bcf_tools = "bcftools"

Execution:

Use the following command to see all the available options in iSAFE.

python2.7 ./src/isafe.py --help

These detailed instructions are also provided in ./help.txt.

Input:

Consider a sample of phased haplotypes in a genomic region. We assume that all sites are biallelic and polymorphic in the sample. Thus, our input is in the form of a binary SNP matrix with each column corresponding to a haplotype and each row to a mutation, and entries corresponding to the allelic state, with 0 denoting the ancestral allele, and 1 denoting the derived allele. Not surprisingly, iSAFE performance deteriorated when the favored mutation is fixed or near fixation (favored allele frequency (ν) > 0.9 in Supplementary Fig. 3e). To handle this special case, we included individuals from non-target populations, using a specific protocol (See online Methods, section Adding outgroup samples). iSAFE can take input in hap or vcf formats.

• hap format: --format hap or -f hap. With hap format, iSAFE assumes:
  • Derived allele is 1 and ancestral allele is 0 in the input file.
  • The selection is ongoing (the favored mutation is not fixed). In the current implementation of iSAFE with --format hap, user is required to add outgroup samples to the input hap file if needed, based on this simple protocol mentioned above.
  • Consequently, input with hap format is not allowed with any of these: --vcf-cont, --sample-case, --sample-cont, --AA.
• vcf format (Default): --format vcf or -f vcf
  • Only phased vcf files are accepted.
  • In v1.0.1 or above it keeps not SNPs (indels, MNPs, and etc) as long as they have only one ALT (binary alleles).
  • vcf format only accepts indexed bgzipped VCF file (.vcf.gz along with tabix index file .vcf.gz.tbi).
  • The Ancestral Allele file (--AA) must be provided with --format vcf.
  • You can choose a subset of samples in the input vcf file by using --sample-case. Otherwise all the samples in the input vcf file are considered as the case samples. See sample ID file format.
  • --vcf-cont is optional but recommended for capturing fixed sweeps. You can choose a subset of samples in this file by using --sample-cont option, otherwise all the samples in this file are cosidered as control population. See sample ID file format.
  • You must use --sample-case and --sample-cont when the --input and --vcf-cont are the same (all samples are provided in a single vcf file).

Note: The software in vcf mode is more flexible and has more options. But if you don't have out-group samples (--vcf-cont is not set), and hap file and vcf file contain the exact same information (iSAFE only cares about position, haplotype phase, derived allele (1), and ancestral allele (0) in the vcf mode), then the output of iSAFE must be identical.

Output:

The output is a non-negative iSAFE-score for each mutation, according to its likelihood of being the favored variant of the selective sweep. Result (<output>.iSAFE.out) is a TAB separated file in the following format.

POS	iSAFE	DAF
291	0.02	0.05
626	0.01	0.55
...	...	...

With following headers:

• POS: Position (bp) sorted in ascending order
• iSAFE: Non-negative iSAFE score
• DAF: Derived allele frequency

Data availability for vcf format

• Download Homo-Sapiens Ancestral Allele files in case you are using --format vcf and consequently --AA:
  • Download links:
    • GRCh37/hg19
    • GRCh38/hg38
  • You need to unzip the files.
• The 1000 Genome Project phased vcf files can be used as --input or --vcf-cont:
  • Download links:
    • GRCh37/hg19
    • GRCh38/hg38
  • You can use your own data and you don't have to use 1000GP data as --input (case) or --vcf-cont (control).
  • Only accept bgzipped vcf files .vcf.gz for faster pre-processing.
  • The tabix index file .vcf.gz.tbi is also required by bcftools along with bgzipped vcf files .vcf.gz.

Demo 1: input in hap format

With --format hap, iSAFE assumes that derived allele is 1 and ancestral allele is 0 in the input file, and the selection is ongoing (the favored mutation is not fixed).

python2.7 ./src/isafe.py --input ./example/hap/demo.hap --output ./example/hap/demo --format hap

• 5Mbp region simulated by msms.
• Position of the favored mutation is 2,500,000.

Demo 2: input in vcf format

Follow the instructions in the Data Requirements section and download Homo-Sapiens Ancestral Allele files and phased vcf files of Chromosome 2 of 1000GP populations (GRCh37/hg19), and replace the text in each < > with the proper file path.

Scenario 1: All samples

All the samples of the --input vcf file as the case population:

• The following command apply iSAFE on 5Mbp region around LCT/MCM6 locus in all 2504 samples (5008 haplotypes) of 1000GP as the case population.

python2.7 ./src/isafe.py --input <chr2 vcf file> --output ./example/vcf/LCT --region 2:134108646-139108646 --AA <chr2 Ancestral Allele file>

Scenario 2: A subset of samples

A subset of samples (--sample-case) of the --input vcf file as the case population:

• The following command apply iSAFE on 5Mbp region around LCT/MCM6 locus in FIN population of 1000GP as the case population.

python2.7 ./src/isafe.py --input <chr2 vcf file> --output ./example/vcf/LCT --region 2:134108646-139108646 --AA <chr2 Ancestral Allele file> --sample-case ./example/vcf/case.sample

Scenario 3: Adding outgroup samples

A subset of samples (--sample-case) of the --input vcf file as the case population and a subset of samples (--sample-cont) of the --vcf-cont vcf file as the control population:

• The following command apply iSAFE on 5Mbp region around LCT/MCM6 locus in FIN population of 1000GP as the case population and YRI, CHB, PEL, and GIH as the control populations.

python2.7 ./src/isafe.py --input <chr2 vcf file> --output ./example/vcf/LCT --region 2:134108646-139108646 --AA <chr2 Ancestral Allele file> --vcf-cont <chr2 vcf file> --sample-case ./example/vcf/case.sample --sample-cont ./example/vcf/cont.sample

• --input and --vcf-cont can point to the same vcf file or different ones. In case they are the same, --sampe-case and --sample-cont are mandatory.
• Position of the putative favored mutation in the FIN population for the selective sweep around LCT/MCM6 locus is 136,608,646 (GRCh37/hg19) of chromosome 2.