jeremymcrae/mupit

De novo mutation recurrence significance testing

Mupit: de novo mutation recurrence significance testing

program to calculate the significance of seeing N DNMs of a specific combination of functional types in a particular gene in M trios

RATIONALE: use gene coding sequence to predict rate of DNMs in coding sequence for each gene of different functional classes, then estimate the probability of seeing the observed combination of different functional classes of DNMs assuming number of DNMs in each class is Poisson distributed

initial implementation: use genome-wide mutation rate and scale by length of coding sequence, use genome-wide average of functional consequences of coding mutations from Kryukov et al. 2007.

Install

pip install git+git://github.com/jeremymcrae/mupit.git --user

# Alternatively:
git clone https://github.com/jeremymcrae/mupit.git
cd mupit
python setup.py install --user

Usage (in python):

from mupit.gene_enrichment import analyse_enrichment
import pandas

trios = {"female": 0, "male": 1}
de_novos = pandas.DataFrame({
    "person_id": ["temp"],
    "chrom": ["16"],
    "start_pos": [89348744],
    "end_pos": [89348744],
    "ref_allele": ["A"],
    "alt_allele": ["G"],
    "hgnc": ["ANKRD11"],
    "consequence": ["missense_variant"],
    "study_code": [None],
    "publication_doi": [None],
    "study_phenotype": [None],
    "type": ["snv"],
    })

enrichment = analyse_enrichment(de_novos, trios)

You can also pass in in a user-specified set of log10-scaled gene-based mutation rates with a dataframe such as:

hgnc	length	mis	non	splice_site	syn	frameshift
AADAC	4364	-4.98	-6.26	-6.74	-5.37	-6.08
ARID1B	6869	-4.06	-5.42	-6.06	-4.30	-5.33
KMT2A	11918	-3.95	-5.12	-5.80	-4.34	-5.09
SCN2A	6109	-4.26	-5.47	-5.30	-4.69	-5.38

The columns are HGNC symbol, CDS length, rate of missense mutations, rate of nonsense mutations, rate of canonical splice site mutations, rate of synonymous mutations, rate of frameshift mutations. The header needs to use the column names given in the table above (hgnc, length, mis, non, css, syn, frameshift).

The gene rates can be used with:

rates = pandas.DataFrame({
    "hgnc": ["ANKRD11"],
    "chrom": ["16"],
    "mis": [9e-5],
    "non": [1e-6],
    "splice_site": [1e-7],
    "syn": [1e-5],
    "frameshift": [1e-6],
    })

enrichment = analyse_enrichment(de_novos, trios, rates=rates)

Format

Read in table of genes and the numbers of families with SNV DNMs within different functional classes and indel DNMS in different functional classes, output same table with added column of probability of seeing that combination of DNMs.

Input data

validated de novo mutations in TSV format:

proband ID | sex | chrom | start | stop | ref | alt | HGNC symbol | VEP consequence | study | DOI | phenotype | SNV or indel

note: current test input file is numbers of mutations, not number of families, some families have >1 mutation in the same gene