BED file parsing

.gitignore

@@ -14,7 +14,7 @@ venv
 result
 .DS_Store
 .env
-data
+data/*
 
 # Logs
 logs
@@ -42,3 +42,4 @@ dist-ssr
 *.sw?
 
 *storybook.log
+src_new/src/core/network_stuff.py

scripts/repeats_to_tsv.py

@@ -0,0 +1,43 @@
+#!/usr/bin/env python3
+
+import sys
+
+import pandas as pd
+
+
+def run(fn):
+    header = []
+    rows = []
+    with open(fn) as fh:
+        for idx, line in enumerate(fh):
+            # print(line)
+            if not header:
+                _ = line.split()
+                header = [
+                    "score",
+                    "div",
+                    "del",
+                    "ins",
+                    "sequence",
+                    "qstart",
+                    "qend",
+                    "qleft",
+                    "C",
+                    "repeat",
+                    "family",
+                    "mstart",
+                    "mend",
+                    "mleft",
+                    "ID",
+                    "last",
+                ]
+            else:
+                row = line.split()
+                rows.append({h: r for h, r in zip(header, row)})
+            # if idx == 30:
+            #    sys.exit(1)
+    pd.DataFrame.from_dict(rows).to_csv(f"{fn}.tsv", sep="\t", index=False)
+
+
+if __name__ == "__main__":
+    run(sys.argv[1])

src_new/config/logging.json

@@ -0,0 +1,40 @@
+{
+  "version": 1,
+  "disable_existing_loggers": false,
+  "formatters": {
+    "simple": {
+      "format": ""
+    }
+  },
+  "handlers": {
+    "console": {
+      "class": "logging.StreamHandler",
+      "level": "DEBUG",
+      "formatter": "simple",
+      "stream": "ext://sys.stdout"
+    },
+    "logfile": {
+      "class": "logging.FileHandler",
+      "level": "DEBUG",
+      "formatter": "simple",
+      "filename": "kinfin.log"
+    }
+  },
+  "loggers": {
+    "my_module": {
+      "level": "DEBUG",
+      "handlers": [
+        "console",
+        "logfile"
+      ],
+      "propagate": false
+    }
+  },
+  "root": {
+    "level": "DEBUG",
+    "handlers": [
+      "console",
+      "logfile"
+    ]
+  }
+}

src_new/data/.gitignore

@@ -0,0 +1,2 @@
+*
+!.gitignore

src_new/docs/getting-started.md

@@ -0,0 +1,20 @@
+# 0. install requirement (i cleaned them up)
+```
+pip install -r requirements.txt
+```
+
+# 1. uncompress tgz's in test
+
+# 2. run examples
+
+```
+KINFIN="/Users/dom/git/genomehubs-kinfin/src_new"
+
+# leps
+OUTDIR="/Users/dom/data/leps"
+python $KINFIN/src/main.py analysis -g $KINFIN/test/leps.orthogroups.txt  -c $KINFIN/test/leps.config.csv -p 10 -v -I -N -P -d $OUTDIR
+
+# nems
+OUTDIR="/Users/dom/data/nems"
+python $KINFIN/src/main.py analysis -g $KINFIN/test/nematodes.orthogroups.txt  -c $KINFIN/test/nematodes.config.csv -i $KINFIN/test/nematodes.interproscan.tsv -f $KINFIN/test/nematodes_fastas -t $KINFIN/test/nematodes.tree.nwk -o CELEG CBRIG -p 10 -v -N -d $OUTDIR
+```

src_new/docs/readme.md

@@ -0,0 +1,21 @@
+### git
+$ git clone https://github.com/genomehubs/kinfin.git
+$ cd genomehubs-kinfin/
+$ git checkout new_cli
+$ cd genomehubs-kinfin/src_new
+
+### create pyenv (or uv)
+$ pyenv virtualenv kinfin
+
+### install dependencies
+$ pip install -r genomehubs-kinfin/src_new/requirements.txt
+
+### edit config file 
+# missing value = NA
+
+### analysis
+### adjust paths to your setup
+$ python genomehubs-kinfin/src_new/src/main.py analysis -g orthogroups.txt -c config.csv -f fastas/ -d psyche_output -X -r family -p 10 -N -v
+
+### convert (feather to tsv)
+$ python genomehubs-kinfin/src_new/src/main.py convert -t FILE.feather -F tsv

src_new/requirements.txt

@@ -0,0 +1,7 @@
+ete4==4.4.0
+pandas==3.0.2
+prettytable==3.17.0
+requests==2.33.0
+scipy==1.13.1
+tqdm==4.67.3
+pyarrow==24.0.0