From d96f82b1a404f5958cc9d25a84ac9da599b5c674 Mon Sep 17 00:00:00 2001
From: Brent Pedersen <bpederse@gmail.com>
Date: Sat, 8 Jun 2024 17:08:50 -0700
Subject: [PATCH] work on csq

---
 Cargo.toml             | 1 +
 scripts/csq.lua        | 8 ++++----
 scripts/csq_header.lua | 8 ++++++++
 scripts/h.lua          | 1 +
 scripts/pre.lua        | 4 ++++
 scripts/t.sh           | 1 +
 6 files changed, 19 insertions(+), 4 deletions(-)
 create mode 100644 scripts/h.lua
 create mode 100644 scripts/pre.lua
 create mode 100644 scripts/t.sh

diff --git a/Cargo.toml b/Cargo.toml
index 36df474..2fd3954 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -13,6 +13,7 @@ env_logger = "0.11.3"
 log = "0.4.21"
 parking_lot = { version = "0.12.1", features = ["arc_lock"] }
 #libc = "0.2.153"
+libc = "0.2"
 
 [[bin]]
 name = "vcfexpr"
diff --git a/scripts/csq.lua b/scripts/csq.lua
index 478bfe3..bc5bb4b 100644
--- a/scripts/csq.lua
+++ b/scripts/csq.lua
@@ -8,8 +8,8 @@ function CSQ.new(fields, header)
     end
     -- now fill a table with keys from header and values from fields
     local self = setmetatable({}, CSQ)
-    for i, field in ipairs(fields) do
-        self[header[i]] = field
+    for i, h in ipairs(header) do
+        self[h] = fields[i]
     end
     return self
 end
@@ -36,12 +36,12 @@ NUMBER_FIELDS = { "AF", "AFR_AF", "AMR_AF", "ASN_AF", "EUR_AF", "EAS_AF", "SAS_A
     "gnomAD_NFE_TSI_AF", "gnomAD_NFE_FOE_AF", "gnomAD_NFE_NWE_AF", "gnomAD_NFE_SEU_AF", "gnomAD_NFE_SWE_AF",
     "gnomAD_NFE_ONF_AF", "gnomAD_NFE_EST_AF", "gnomAD_NFE_MED_AF",
     "gnomAD_NFE_SCA_AF", "gnomAD_NFE_BAL_AF", "gnomAD_NFE_IB",
-    "MAX_AF", "MAX_AF_POPS", }
+    "MAX_AF", "MAX_AF_POPS", "ALLELE_NUM", "DISTANCE"
+}
 -- if the field starts with gnomAD_ also add gnomADe_... and gnomADg_...
 add_gnomad = {}
 for _, field in ipairs(NUMBER_FIELDS) do
     if string.match(field, "^gnomAD_") then
-        print(string.gsub(field, "^gnomAD_", "gnomADe_"))
         add_gnomad[#add_gnomad + 1] = string.gsub(field, "^gnomAD_", "gnomADe_")
         add_gnomad[#add_gnomad + 1] = string.gsub(field, "^gnomAD_", "gnomADg_")
     end
diff --git a/scripts/csq_header.lua b/scripts/csq_header.lua
index 16c89b7..ffd4825 100644
--- a/scripts/csq_header.lua
+++ b/scripts/csq_header.lua
@@ -57,4 +57,12 @@ for i, v in ipairs(parsed_table3) do
     print(i, v)
 end
 
+
+desc =
+'Functional annotations: \'Allele | Annotation | Annotation_Impact | Gene_Name | Gene_ID | Feature_Type | Feature_ID | Transcript_BioType | Rank | HGVS.c | HGVS.p | cDNA.pos / cDNA.length | CDS.pos / CDS.length | AA.pos / AA.length | Distance | ERRORS / WARNINGS / INFO\'"'
+
+for i, v in ipairs(parse_description(desc)) do
+    print(i, v)
+end
+
 --]]
diff --git a/scripts/h.lua b/scripts/h.lua
new file mode 100644
index 0000000..50d2da7
--- /dev/null
+++ b/scripts/h.lua
@@ -0,0 +1 @@
+desc = parse_description(header_desc)
diff --git a/scripts/pre.lua b/scripts/pre.lua
new file mode 100644
index 0000000..e645f6b
--- /dev/null
+++ b/scripts/pre.lua
@@ -0,0 +1,4 @@
+header:add_info({ID="af_copy", Number=1, Description="adding a single field", Type="Float"})
+--print(header:info_get("af_copy").Description)
+
+header_desc = header:info_get("vep").Description
diff --git a/scripts/t.sh b/scripts/t.sh
new file mode 100644
index 0000000..8f9dddb
--- /dev/null
+++ b/scripts/t.sh
@@ -0,0 +1 @@
+./target/debug/vcfexpr filter -e "print(CSQ.new(variant:info('vep', 0), desc).Allele); return variant.id == 'rs2124717267'"  -o var.bcf gnomad.genomes.v4.0.sites.chrY.vcf.bgz -p scripts/pre.lua -s 'af_copy=return variant:info("AF", 0)' -l scripts/csq.lua -l scripts/csq_header.lua   -l  scripts/h.lua