FAQ.md

Variable pool sizes

The input file (--bams) to CRISP should specify the pool size (2 x number of individuals, assuming diploid genomes) on each line as follows:

pool1.bam PS=50

pool2.bam PS=40

...

pool20.bam PS=36

CRISP can run in two modes: --EM 0 and --EM 1 option

--EM 1 corresponds to the new CRISP method that can estimate the 'discrete' pooled genotype for each pool and uses a likelihood ratio test (in addition to the contingency table analysis) to call variants. This is recommended for artificial pooled sequencing studies.

--EM 0 corresponds to the original CRISP method that uses the contingency table analysis and per pool filters to identify variant sites and report allele frequencies for each pool. The ctpval and qvpval thresholds are only used with EM = 0 option

running CRISP on targeted sequencing experiments

It is recommended to specify a bedfile (--bed option) for targeted sequencing experiments. If a bedfile is not specified, the program will evaluate each base for variant calling and the output file can be quite large. If the bedfile is specified, CRISP will only call variants in the regions listed in the bedfile. To call variants in the region flanking the targeted intervals, --flanking option can be used. With "--flanking 50" option, variants will also be called in a 50 bp region flanking every interval.

Some requirements before running CRISP

1. The reference sequence file should be indexed using 'samtools faidx' or a similar program and placed in same directory as fasta file with extension .fai
2. The aligned reads for each pool should be in a single bam file that is sorted by chromosomal coordinates. If there are multiple bam files for each pool, merge them using samtools or Picard.
3. BAM files should be indexed in order to use the --regions option with the indexed bam file named as pooln.bam.bai (pooln.bai will not work)
4. Please make sure that the reference sequence file is the same as the one used to align the reads in the BAM files

calling of short insertions/deletions (indels)

CRISP can call short indels as well as SNVs. However, it is difficult to call multi-allelic indels ( i.e. indels with multiple variant alleles) from pooled sequence data. For indel analysis, CRISP assumes that indels are left justified, --leftalign 1 option can be used to left justify gaps in aligned reads

Using the CRISP VCF with other tools

1. The CRISP VCF should be converted to a standard pooled-genotype VCF using the python script "convert_pooled_vcf.py" available in the scripts sub-folder

Empty VCF file after running CRISP

This can be caused by:

1. Low mapping quality of reads (default filter of 20)
2. Chromosome names in the bed file are different the reference genome
3. Issues with the bed file (try running CRISP using the --regions option on a small genomic region)
4. using the option "--qvoffset 64" when the quality value offset is 33