deleted: cre.gnomad_genome.conf

	new file:   cre.vcfanno.conf
	new file:   cre.vcfanno.lua

cre.vcfanno.conf

@@ -0,0 +1,208 @@
+[[annotation]]
+file="gnomad.exomes.r2.0.1.sites.no-VEP.nohist.tidy.vcf.gz"
+fields=["AF", "AC", "Hom"]
+names=["Gnomad_af_es", "Gnomad_ac_es", "Gnomad_hom_es"]
+ops=["self", "self", "self"]
+
+[[annotation]]
+file="variation/gnomad_genome.vcf.gz"
+fields=["AF", "AC", "Hom"]
+names=["Gnomad_af_gs", "Gnomad_ac_gs", "Gnomad_hom_gs"]
+ops=["self", "self", "self"]
+
+[[annotation]]
+file="ExAC.r0.3.sites.vep.tidy.vcf.gz"
+fields = ["AF","AC","AC_Het", "AC_Hom"]
+ names = ["Exac_af","Exac_ac","Exac_ac_het", "Exac_hom"]
+ops=["self","self","self","self"] 
+
+[[annotation]]
+file="ESP6500SI.all.snps_indels.tidy.v2.vcf.gz"
+fields=["EA_AC", "AA_AC", "TAC"]
+names=["af_esp_ea_float", "af_esp_aa_float", "af_esp_all_float"]
+ops=["lua:ratio(vals)", "lua:ratio(vals)", "lua:ratio(vals)"]
+
+
+[[annotation]]
+file="dbsnp.b147.20160601.tidy.vcf.gz"
+fields=["ID"]
+names=["rs_ids"]
+ops=["concat"]
+
+[[annotation]]
+file="ALL.wgs.phase3_shapeit2_mvncall_integrated_v5a.20130502.sites.tidy.vcf.gz"
+fields=["AMR_AF", "EAS_AF", "SAS_AF", "AFR_AF", "EUR_AF", "AF"]
+names=["af_1kg_amr", "af_1kg_eas", "af_1kg_sas", "af_1kg_afr", "af_1kg_eur", "af_1kg_all"]
+ops=["max", "max", "max", "max", "max", "max"]
+
+
+[[annotation]]
+file="clinvar_20170130.tidy.vcf.gz"
+fields=["CLNSIG", "CLNDBN"]
+names=["clinvar_pathogenic", "clinvar_disease_name"]
+ops=["self", "self"]
+
+# convert 5 to 'pathogenic', 255 to 'unknown', etc.
+[[postannotation]]
+fields=["clinvar_pathogenic"]
+op="lua:clinvar_sig(clinvar_pathogenic)"
+name="clinvar_sig"
+type="String"
+
+# calculate allele frequencies for all populations.
+[[postannotation]]
+fields=["ac_exac_all", "an_exac_all"]
+name="af_exac_all"
+op="div2"
+type="Float"
+
+                    [[postannotation]]
+                    fields=["ac_adj_exac_afr", "an_adj_exac_afr"]
+                    name="af_adj_exac_afr"
+                    op="div2"
+                    type="Float"
+
+                    [[postannotation]]
+                    fields=["ac_adj_exac_amr", "an_adj_exac_amr"]
+                    name="af_adj_exac_amr"
+                    op="div2"
+                    type="Float"
+
+                    [[postannotation]]
+                    fields=["ac_adj_exac_eas", "an_adj_exac_eas"]
+                    name="af_adj_exac_eas"
+                    op="div2"
+                    type="Float"
+
+                    [[postannotation]]
+                    fields=["ac_adj_exac_fin", "an_adj_exac_fin"]
+                    name="af_adj_exac_fin"
+                    op="div2"
+                    type="Float"
+
+                    [[postannotation]]
+                    fields=["ac_adj_exac_nfe", "an_adj_exac_nfe"]
+                    name="af_adj_exac_nfe"
+                    op="div2"
+                    type="Float"
+
+                    [[postannotation]]
+                    fields=["ac_adj_exac_oth", "an_adj_exac_oth"]
+                    name="af_adj_exac_oth"
+                    op="div2"
+                    type="Float"
+
+                    [[postannotation]]
+                    fields=["ac_adj_exac_sas", "an_adj_exac_sas"]
+                    name="af_adj_exac_sas"
+                    op="div2"
+                    type="Float"
+
+
+                    [[postannotation]]
+                    fields=['af_adj_exac_afr', 'af_adj_exac_amr', 'af_adj_exac_eas', 'af_adj_exac_fin', 'af_adj_exac_nfe', 'af_adj_exac_oth', 'af_adj_exac_sas', "af_esp_ea", "af_esp_aa", "af_esp_all", "af_1kg_amr", "af_1kg_eas", "af_1kg_sas", "af_1kg_afr", "af_1kg_eur", "af_1kg_all"]
+                    op="max"
+                    name="max_aaf_all"
+                    type="Float"
+
+                    [[postannotation]]
+                    fields=["clinvar_sig", "max_aaf_all"]
+                    op="lua:check_clinvar_aaf(clinvar_sig, max_aaf_all, 0.005)"
+                    name="common_pathogenic"
+                    type="Flag"
+
+                    [[annotation]]
+                    file="cosmic-v68-GRCh37.tidy.vcf.gz"
+                    fields=["ID"]
+                    names=["cosmic_ids"]
+                    ops=["uniq"]
+
+
+                    [[annotation]]
+                    file="hg19_fitcons_fc-i6-0_V1-01.bed.gz"
+                    columns=[4]
+                    names=["fitcons_float"]
+                    ops=["mean"]
+
+
+
+                    #[[annotation]]
+                    #file="LCR-hs37d5.bed.gz"
+                    #names=["LCR"]
+                    #columns=[2]
+                    #ops=["flag"]
+                    #
+                    #[[annotation]]
+                    #http://humanparalogy.gs.washington.edu/build37/build37.htm
+                    #  wget -O - http://humanparalogy.gs.washington.edu/build37/data/GRCh37GenomicSuperDup.tab \
+                    #      | tail -n+2 \
+                    #      | grep -Pv "_gl00|_random|chrUn" | sort -k1,1V -k2,2n | bgzip -c > GRCh37GenomicSuperDup.tab.gz
+                    #file="GRCh37GenomicSuperDup.tab.gz"
+                    #names=["superdup"]
+                    #columns=[2]
+                    #ops=["flag"]
+
+
+
+                    [[annotation]]
+                    file="encode.6celltypes.consensus.bedg.gz"
+                    #chrom   start   end     gm12878 h1hesc  helas3  hepg2   huvec   k562
+                    columns=[4,5,6,7,8,9]
+                    ops=["concat","concat", "concat", "concat", "concat", "concat"]
+                    names=[ "encode_consensus_gm12878", "encode_consensus_h1hesc", "encode_consensus_helas3", "encode_consensus_hepg2", "encode_consensus_huvec", "encode_consensus_k562"]
+
+                    [[annotation]]
+                    file="hg19.gwas.bed.gz"
+                    columns=[4]
+                    names=["gwas_pubmed_trait"]
+                    ops=["uniq"]
+
+                    [[annotation]]
+                    file="hg19.rmsk.bed.gz"
+                    columns=[4]
+                    names=["rmsk"]
+                    ops=["uniq"]
+
+                    [[annotation]]
+                    file="hg19.gerp.elements.bed.gz"
+                    columns=[4]
+                    names=["gerp_elements"]
+                    ops=["mean"]
+
+                    [[annotation]]
+                    file="hg19.CpG.bed.gz"
+                    columns=[2]
+                    names=["cpg_island"]
+                    ops=["flag"]
+
+                    [[annotation]]
+                    file="hg19.dgv.bed.gz"
+                    columns=[4]
+                    names=["dgv"]
+                    ops=["uniq"]
+
+                    [[annotation]]
+                    file="wgEncodeRegTfbsClusteredV2.cell_count.20130213.bed.gz"
+                    columns=[4]
+                    names=["tfbs"]
+                    ops=["uniq"]
+
+                    [[annotation]]
+                    file="genetic_map_HapMapII_GRCh37.gz"
+                    columns=[4, 5]
+                    ops=["mean", "mean"]
+                    names=["hapmap1", "hapmap2"]
+
+
+                    [[annotation]]
+                    file="stam.125cells.dnaseI.hg19.bed.gz"
+                    columns=[5, 6]
+                    ops=["mean", "uniq"]
+                    names=["stam_mean", "stam_names"]
+
+                    [[annotation]]
+                    file="cse-hiseq-8_4-2013-02-20.bed.gz"
+                    columns=[2]
+                    ops=["flag"]
+                    names=["cse-hiseq"]
+