From 202b9e8197c2b02641d7f022f8b2ad068c24624b Mon Sep 17 00:00:00 2001
From: Koshlan Mayer-Blackwell <46639063+kmayerb@users.noreply.github.com>
Date: Thu, 6 May 2021 16:52:14 -0700
Subject: [PATCH] accomodate array in embedDistanceMatrix
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

accomodate array as input argument dmatDF in embedDistanceMatrix()
see issue #2 for explanation




https://github.com/agartland/utils/blob/b61935a860838a0e70afde7c9ecf2c68f51a2c4b/embedding.py#L63-L69

Note that the upper portion of the function is meant to accomodate either numpy.array or pd.DataFrame passed to dmatDf, but if numpy array is supplied then L111 will raise an error:

https://github.com/agartland/utils/blob/b61935a860838a0e70afde7c9ecf2c68f51a2c4b/embedding.py#L111


Example
```python
embedDistanceMatrix(dmatDf = pd.DataFrame(tr.pw_beta[0:10,:][:,0:10]), method = "mds")
```
👍

```python
embedDistanceMatrix(dmatDf = tr.pw_beta[0:10,:][:,0:10], method = "mds")
```
👎
```
AttributeError: 'numpy.ndarray' object has no attribute 'index'
```
---
 embedding.py | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

diff --git a/embedding.py b/embedding.py
index 73e4f80..5201a9e 100644
--- a/embedding.py
+++ b/embedding.py
@@ -108,7 +108,12 @@ def embedDistanceMatrix(dmatDf, method='kpca', n_components=2, **kwargs):
         return
 
     assert xy.shape[0] == dmatDf.shape[0]
-    xyDf = pd.DataFrame(xy[:, :n_components], index=dmatDf.index, columns=np.arange(n_components))
+    
+    if isinstance(dmatDF, pd.DataFrame):
+        xyDf = pd.DataFrame(xy[:, :n_components], index=dmatDf.index, columns=np.arange(n_components))
+    else:
+        xyDf = pd.DataFrame(xy[:, :n_components], columns=np.arange(n_components))
+        
     if method == 'kpca':
         """Not sure how negative eigenvalues should be handled here, but they are usually
         small so it shouldn't make a big difference"""