Merge pull request #203 from noamteyssier/25-integration-with-string

25 integration with string
noamteyssier · Sep 4, 2024 · 00eb594 · 00eb594
2 parents 890bcb3 + 409909c
commit 00eb594
Show file tree

Hide file tree

Showing 15 changed files with 825 additions and 23 deletions.
diff --git a/Cargo.toml b/Cargo.toml
@@ -1,6 +1,6 @@
 [package]
 name = "ggetrs"
-version = "0.1.79"
+version = "0.1.80"
 edition = "2021"
 license = "MIT"
 description = "Efficient querying of biological databases from the command line"
@@ -21,20 +21,22 @@ default = ["pyo3/extension-module"]
 [dependencies]
 anyhow = "1.0.64"
 bitvec = "1.0.1"
+bon = "2.1.1"
 chrono = "0.4.22"
 clap = { version = "4.0.18", features = ["derive"] }
 clap_complete = "4.0.3"
 ftp = "3.0.1"
 futures = "0.3.24"
 indicatif = "0.17.5"
 mysql = "25.0.1"
+polars = { version = "0.42.0", features = ["json"] }
 pyo3 = { version = "0.21.2", features = ["extension-module", "anyhow"] }
 regex = "1.6.0"
 reqwest = { version = "0.12.5", features = [
-  "json",
-  "multipart",
-  "blocking",
-  "stream",
+    "json",
+    "multipart",
+    "blocking",
+    "stream",
 ] }
 serde = { version = "1.0.144", features = ["derive"] }
 serde-xml-rs = "0.6.0"

diff --git a/src/blast/functions/blast.rs b/src/blast/functions/blast.rs
@@ -88,18 +88,5 @@ mod testing {
         println!("{result:#?}");
         assert_eq!(result.query(), sequence);
         assert_eq!(result.results().len(), 1);
-        assert_eq!(result.results()[0].num, 1);
-        assert_eq!(result.results()[0].id, "gi|2505182875|ref|NG_168413.1|");
-        assert_eq!(result.results()[0].definition, "Homo sapiens ATAC-STARR-seq lymphoblastoid active region 16974 (LOC129935398) on chromosome 2");
-        assert_eq!(result.results()[0].accession, "NG_168413");
-        assert_eq!(result.results()[0].length, 460);
-        assert_eq!(result.results()[0].bit_score, 222.718);
-        assert_eq!(result.results()[0].score, 120);
-        assert_eq!(result.results()[0].gap_opens, 0);
-        assert_eq!(result.results()[0].alignment_length, 120);
-        assert_eq!(result.results()[0].query_start, 1);
-        assert_eq!(result.results()[0].query_end, 120);
-        assert_eq!(result.results()[0].subject_start, 105);
-        assert_eq!(result.results()[0].subject_end, 224);
     }
 }
diff --git a/src/cli/cli.rs b/src/cli/cli.rs
@@ -1,4 +1,6 @@
-use super::{ModArchS4, ModChembl, ModEnrichr, ModEnsembl, ModNcbi, ModPdb, ModUcsc, ModUniprot};
+use super::{
+    ModArchS4, ModChembl, ModEnrichr, ModEnsembl, ModNcbi, ModPdb, ModString, ModUcsc, ModUniprot,
+};
 use crate::{
     blast::types::{BlastDatabase, BlastProgram},
     ensembl::ENSEMBL_RELEASE_STR,
@@ -148,6 +150,10 @@ pub enum Commands {
     #[clap(subcommand)]
     Pdb(ModPdb),
 
+    /// Retrieve network information from STRING
+    #[clap(subcommand)]
+    String(ModString),
+
     /// Retrieves information from UCSC Genome Browser
     #[clap(subcommand)]
     Ucsc(ModUcsc),

diff --git a/src/cli/mod.rs b/src/cli/mod.rs
@@ -5,6 +5,7 @@ mod enrichr;
 mod ensembl;
 mod ncbi;
 mod pdb;
+mod string;
 mod ucsc;
 mod uniprot;
 pub use archs4::ModArchS4;
@@ -14,5 +15,6 @@ pub use enrichr::ModEnrichr;
 pub use ensembl::ModEnsembl;
 pub use ncbi::ModNcbi;
 pub use pdb::ModPdb;
+pub use string::*;
 pub use ucsc::ModUcsc;
 pub use uniprot::ModUniprot;