Add ht schema generation workflow

data-pipeline/schemas/gnomad_v4_variants/prepare_gnomad_v4_exome_variants/input_path/mock_v4_release.ht.schema

@@ -0,0 +1,338 @@
+----------------------------------------
+Global fields:
+    'freq_meta': array<dict<str, str>> 
+    'freq_index_dict': dict<str, int32> 
+    'faf_meta': array<dict<str, str>> 
+    'faf_index_dict': dict<str, int32> 
+    'freq_sample_count': array<int32> 
+    'filtering_model': struct {
+        model_name: str, 
+        score_name: str, 
+        feature_medians: dict<tuple (
+            str
+        ), struct {
+            a_index: int32, 
+            n_alt_alleles: int32, 
+            AS_pab_max: float64, 
+            AS_MQRankSum: float64, 
+            AS_SOR: float64, 
+            AS_ReadPosRankSum: float64
+        }>, 
+        variants_by_strata: dict<tuple (
+            str
+        ), int64>, 
+        features_importance: dict<str, float64>, 
+        features: array<str>, 
+        test_results: array<struct {
+            rf_prediction: str, 
+            rf_label: str, 
+            n: int32
+        }>, 
+        rf_snv_cutoff: struct {
+            bin: float64, 
+            min_score: float64
+        }, 
+        rf_indel_cutoff: struct {
+            bin: float64, 
+            min_score: float64
+        }, 
+        inbreeding_cutoff: float64, 
+        model_id: str
+    } 
+    'tool_versions': struct {
+        dbsnp_version: str, 
+        cadd_version: str, 
+        revel_version: str, 
+        splicaai_version: str, 
+        primateai_version: str, 
+        pangolin_version: str, 
+        vrs_version: str
+    } 
+    'vep_globals': struct {
+        vep_version: str, 
+        vep_csq_header: str, 
+        vep_help: str, 
+        vep_config: str
+    } 
+    'age_distribution': struct {
+        bin_edges: array<float64>, 
+        bin_freq: array<int32>, 
+        n_smaller: int32, 
+        n_larger: int32
+    } 
+    'age_index_dict': dict<str, int32> 
+    'age_meta': array<dict<str, str>> 
+    'grpmax_index_dict': dict<str, int32> 
+    'grpmax_meta': array<dict<str, str>> 
+    'README': dict<str, str> 
+    'gnomad_qc_repo': str 
+    'gnomad_methods_repo': str 
+----------------------------------------
+Row fields:
+    'locus': locus<GRCh38> 
+    'alleles': array<str> 
+    'freq': array<struct {
+        AC: int32, 
+        AF: float64, 
+        AN: int32, 
+        homozygote_count: int32
+    }> 
+    'grpmax': array<struct {
+        AC: int32, 
+        AF: float64, 
+        AN: int32, 
+        homozygote_count: int32, 
+        grp: str, 
+        faf95: float64
+    }> 
+    'faf': array<struct {
+        faf95: float64, 
+        faf99: float64
+    }> 
+    'a_index': int32 
+    'was_split': bool 
+    'rsid': set<str> 
+    'filters': set<str> 
+    'info': struct {
+        QUALapprox: int64, 
+        SB: array<int32>, 
+        MQ: float64, 
+        MQRankSum: float64, 
+        VarDP: int32, 
+        AS_ReadPosRankSum: float64, 
+        AS_pab_max: float64, 
+        AS_QD: float32, 
+        AS_MQ: float64, 
+        QD: float32, 
+        AS_MQRankSum: float64, 
+        FS: float64, 
+        AS_FS: float64, 
+        ReadPosRankSum: float64, 
+        AS_QUALapprox: int64, 
+        AS_SB_TABLE: array<int32>, 
+        AS_VarDP: int32, 
+        AS_SOR: float64, 
+        SOR: float64, 
+        singleton: bool, 
+        transmitted_singleton: bool, 
+        omni: bool, 
+        mills: bool, 
+        monoallelic: bool, 
+        AS_VQSLOD: float64, 
+        InbreedingCoeff: float64, 
+        vrs: struct {
+            VRS_Allele_IDs: array<str>, 
+            VRS_Starts: array<int32>, 
+            VRS_Ends: array<int32>, 
+            VRS_States: array<str>
+        }
+    } 
+    'vep': struct {
+        allele_string: str, 
+        end: int32, 
+        id: str, 
+        input: str, 
+        intergenic_consequences: array<struct {
+            allele_num: int32, 
+            consequence_terms: array<str>, 
+            impact: str, 
+            variant_allele: str
+        }>, 
+        most_severe_consequence: str, 
+        motif_feature_consequences: array<struct {
+            allele_num: int32, 
+            consequence_terms: array<str>, 
+            high_inf_pos: str, 
+            impact: str, 
+            motif_feature_id: str, 
+            motif_name: str, 
+            motif_pos: int32, 
+            motif_score_change: float64, 
+            transcription_factors: array<str>, 
+            strand: int32, 
+            variant_allele: str
+        }>, 
+        regulatory_feature_consequences: array<struct {
+            allele_num: int32, 
+            biotype: str, 
+            consequence_terms: array<str>, 
+            impact: str, 
+            regulatory_feature_id: str, 
+            variant_allele: str
+        }>, 
+        seq_region_name: str, 
+        start: int32, 
+        strand: int32, 
+        transcript_consequences: array<struct {
+            allele_num: int32, 
+            amino_acids: str, 
+            appris: str, 
+            biotype: str, 
+            canonical: int32, 
+            ccds: str, 
+            cdna_start: int32, 
+            cdna_end: int32, 
+            cds_end: int32, 
+            cds_start: int32, 
+            codons: str, 
+            consequence_terms: array<str>, 
+            distance: int32, 
+            domains: array<struct {
+                db: str, 
+                name: str
+            }>, 
+            exon: str, 
+            flags: str, 
+            gene_id: str, 
+            gene_pheno: int32, 
+            gene_symbol: str, 
+            gene_symbol_source: str, 
+            hgnc_id: str, 
+            hgvsc: str, 
+            hgvsp: str, 
+            hgvs_offset: int32, 
+            impact: str, 
+            intron: str, 
+            lof: str, 
+            lof_flags: str, 
+            lof_filter: str, 
+            lof_info: str, 
+            mane_select: str, 
+            mane_plus_clinical: str, 
+            mirna: array<str>, 
+            polyphen_prediction: str, 
+            polyphen_score: float64, 
+            protein_end: int32, 
+            protein_start: int32, 
+            protein_id: str, 
+            sift_prediction: str, 
+            sift_score: float64, 
+            source: str, 
+            strand: int32, 
+            transcript_id: str, 
+            tsl: int32, 
+            uniprot_isoform: array<str>, 
+            variant_allele: str
+        }>, 
+        variant_class: str
+    } 
+    'rf': struct {
+        rf_positive_label: bool, 
+        rf_negative_label: bool, 
+        rf_label: str, 
+        rf_train: bool, 
+        rf_tp_probability: float64
+    } 
+    'region_flag': struct {
+        lcr: bool, 
+        segdup: bool, 
+        non_par: bool
+    } 
+    'allele_info': struct {
+        variant_type: str, 
+        allele_type: str, 
+        n_alt_alleles: int32, 
+        was_mixed: bool
+    } 
+    'histograms': struct {
+        qual_hists: struct {
+            gq_hist_all: struct {
+                bin_edges: array<float64>, 
+                bin_freq: array<int64>, 
+                n_smaller: int64, 
+                n_larger: int64
+            }, 
+            dp_hist_all: struct {
+                bin_edges: array<float64>, 
+                bin_freq: array<int64>, 
+                n_smaller: int64, 
+                n_larger: int64
+            }, 
+            gq_hist_alt: struct {
+                bin_edges: array<float64>, 
+                bin_freq: array<int64>, 
+                n_smaller: int64, 
+                n_larger: int64
+            }, 
+            dp_hist_alt: struct {
+                bin_edges: array<float64>, 
+                bin_freq: array<int64>, 
+                n_smaller: int64, 
+                n_larger: int64
+            }, 
+            ab_hist_alt: struct {
+                bin_edges: array<float64>, 
+                bin_freq: array<int64>, 
+                n_smaller: int64, 
+                n_larger: int64
+            }
+        }, 
+        raw_qual_hists: struct {
+            gq_hist_all: struct {
+                bin_edges: array<float64>, 
+                bin_freq: array<int64>, 
+                n_smaller: int64, 
+                n_larger: int64
+            }, 
+            dp_hist_all: struct {
+                bin_edges: array<float64>, 
+                bin_freq: array<int64>, 
+                n_smaller: int64, 
+                n_larger: int64
+            }, 
+            gq_hist_alt: struct {
+                bin_edges: array<float64>, 
+                bin_freq: array<int64>, 
+                n_smaller: int64, 
+                n_larger: int64
+            }, 
+            dp_hist_alt: struct {
+                bin_edges: array<float64>, 
+                bin_freq: array<int64>, 
+                n_smaller: int64, 
+                n_larger: int64
+            }, 
+            ab_hist_alt: struct {
+                bin_edges: array<float64>, 
+                bin_freq: array<int64>, 
+                n_smaller: int64, 
+                n_larger: int64
+            }
+        }, 
+        age_hists: array<struct {
+            age_hist_hom: struct {
+                bin_edges: array<float64>, 
+                bin_freq: array<int64>, 
+                n_smaller: int64, 
+                n_larger: int64
+            }, 
+            age_hist_ht: struct {
+                bin_edges: array<float64>, 
+                bin_freq: array<int64>, 
+                n_smaller: int64, 
+                n_larger: int64
+            }
+        }>
+    } 
+    'in_silico_predictors': struct {
+        cadd: struct {
+            phred: float32, 
+            raw_score: float32, 
+            has_duplicate: bool
+        }, 
+        revel: struct {
+            revel_score: float64, 
+            has_duplicate: bool
+        }, 
+        splice_ai: struct {
+            splice_ai_score: float32, 
+            splice_consequence: str, 
+            has_duplicate: bool
+        }, 
+        pangolin: struct {
+            pangolin_score: float64
+        }
+    } 
+----------------------------------------
+Key: ['locus', 'alleles']
+----------------------------------------