• Assigned: Flisso
• type: bug
• status: in progress
• key words: cross, qtl2, calc_genoprob, bugs

Running subset of genotype and founders csv on qtl2 to generate founder aware smoothed genotypes. The script is crushing as per the followin error message:

calc_genoprob failing with negative length vectors are not allowed

For reference, see "qtl2_hmm_pipeline.R" script:

The following were key findings from the run, and the error:

• Map and IDs were consistent:
• - 50,000 markers
• - no duplicate marker IDs
• - monotonic increasing cM
• Genotype dimensions:
• - HS genotypes: 1499 x 50000
• - Founder genotypes: 8 x 50000
• Error cause matched integer-length overflow conditions:
• The original workflow tried to allocate a genotype-probability object effectively sized around 1499 * 50000 * 36 = 2,698,200,000, which exceeds R’s 32-bit vector-length limit (2,147,483,647), causing negative length vectors are not allowed.
• So the solution was to chunk the files to 5000 lines, but still the culprit is on the calc_genoprob() runtime.

• [x] error: "calc_genoprob failing with negative length vectors are not

allowed)"

• [] Re-run the script per specified chunks
• [] Evaluate the smoothed output for its validity and intepretability
• [] use the proximal/distal founder aware markers to extract snps from the original geno file.
• [] or, extend a function in the script to perform this
• [] Test the results with gemma and rqtl2 mapping