phase_common failing at PBWT step

[nmshahir]

I'm using SHAPEIT5 to phase a large panel (300K+) using phase_common and it keeps stopping at the PBWT step?
This is the relevant error:

WARNING: No PBWT states found [1000042 / w=76] / Using 100 random states

WARNING: No PBWT states found [1000042 / w=77] / Using 100 random states

ERROR: Haploid underflow impossible to recover for [1000039]

The parameters used are below:

Parameters:

• Seed : 15052011
• Threads : 8 threads
• MCMC : 15 iterations [5b + 1p + 1b + 1p + 1b + 1p + 5m]
• PBWT : [window = 4cM / depth = auto / modulo = auto / mac = 5 / missing = 0.1]
• HMM : [window = 4cM / Ne = 15000 / Recombination rates given by genetic map]

Please advise?

[Thecover]

Hi,
I encountered the same issue as you. Did you manage to resolve it? If so, could you please share how you did it? Thank you!

[LouisLeNezet]

I've ran into the same issue but it appeared only when adding the genetic map.

wget https://raw.githubusercontent.com/nf-core/test-datasets/phaseimpute/hum_data/panel/chr21/1000GP.chr21.s.norel.vcf.gz
wget https://raw.githubusercontent.com/nf-core/test-datasets/phaseimpute/hum_data/panel/chr21/1000GP.chr21.s.norel.vcf.gz.csi
wget https://raw.githubusercontent.com/nf-core/test-datasets/phaseimpute/hum_data/reference_genome/GRCh38_21.map
SHAPEIT5_phase_common \
        --input 1000GP.chr21.s.norel.vcf.gz \
        --map GRCh38_21.map \
        --region chr21:16570070-16609998 \
        --thread 2 \
        --output 1000GP_chr21_16570070-16609998_chunks.vcf.gz

[sadafrf]

Hi, I found the same issue and I am also using genetic map, did you come up with a solution for this? @LouisLeNezet @Thecover @nmshahir

[LouisLeNezet]

I'm afraid not for the moment.