From 1509b0a4818e7d3e281becbc125a7717725cac95 Mon Sep 17 00:00:00 2001
From: Philipp Schlegel <observing@web.de>
Date: Sun, 10 Jan 2021 16:18:07 +0000
Subject: [PATCH] add FAFB14 mirror registration

---
 flybrains/core.py                          | 22 +++++++++++++++---
 flybrains/data/FAFB14_mirror_landmarks.csv | 27 ++++++++++++++++++++++
 flybrains/data/README.md                   |  7 ++++++
 flybrains/templates.py                     |  4 ++--
 4 files changed, 55 insertions(+), 5 deletions(-)
 create mode 100644 flybrains/data/FAFB14_mirror_landmarks.csv

diff --git a/flybrains/core.py b/flybrains/core.py
index 28d48dd..96e6368 100644
--- a/flybrains/core.py
+++ b/flybrains/core.py
@@ -23,11 +23,17 @@
 from navis import transforms
 
 import numpy as np
+import pandas as pd
 
 from .download import get_data_home
 
 __all__ = ['register_transforms', 'report']
 
+# Read in meta data
+fp = os.path.dirname(__file__)
+
+data_filepath = os.path.join(fp, 'data')
+
 
 @functools.lru_cache()
 def get_nat_regdirs(verbose=False):
@@ -229,10 +235,10 @@ def register_transforms():
                                            transform_type='bridging')
 
     # Add transform between JRCFIB2018F (nm) and JRCFIB2018Fum (um)
-    tr = transforms.affine.AffineTransform(np.diag([1/1e3, 1/1e3, 1/1e3, 1]))
+    tr = transforms.affine.AffineTransform(np.diag([1e3, 1e3, 1e3, 1]))
     transforms.registry.register_transform(transform=tr,
-                                           source='JRCFIB2018F',
-                                           target='JRCFIB2018Fum',
+                                           source='JRCFIB2018Fum',
+                                           target='JRCFIB2018F',
                                            transform_type='bridging')
 
     # Add transform between FAFB14um and FAFB14 (nm)
@@ -255,3 +261,13 @@ def register_transforms():
                                            source='JFRC2',
                                            target='JFRC2010',
                                            transform_type='bridging')
+
+    # Add a simple mirror transform for FAFB14
+    fp = os.path.join(data_filepath, 'FAFB14_mirror_landmarks.csv')
+    lm = pd.read_csv(fp)
+    tr = transforms.thinplate.TPStransform(lm[['x_mirr', 'y_mirr', 'z_mirr']].values,
+                                           lm[['x', 'y', 'z']].values)
+    transforms.registry.register_transform(transform=tr,
+                                           source='FAFB14',
+                                           target=None,
+                                           transform_type='mirror')
diff --git a/flybrains/data/FAFB14_mirror_landmarks.csv b/flybrains/data/FAFB14_mirror_landmarks.csv
new file mode 100644
index 0000000..a56302f
--- /dev/null
+++ b/flybrains/data/FAFB14_mirror_landmarks.csv
@@ -0,0 +1,27 @@
+label,x,y,z,side,x_mirr,y_mirr,z_mirr,original_side
+antler,587201,182007,185345,L,568923,179977,195734,R
+dimple1,697109,222823,58876,L,705480,205422,106682,R
+dimple2,607100,104989,138065,L,586352,98028,150176,R
+dimple3,622471,308005,144728,L,608265,304754,157541,R
+dimple4,643639,214585,66168,L,645039,199521,93013,R
+LH,733031,146563,152677,L,721699,134702,182934,R
+MB1,692958,172890,65884,L,681095,154893,102163,R
+MB2,599003,93341,76730,L,591704,85278,90487,R
+peak,619804,294979,53835,L,630690,275831,66134,R
+SEZ,559790,317818,37401,L,561715,311357,47117,R
+vis1,730150,189507,204076,L,731312,183857,235749,R
+vis2,680242,344243,170947,L,680339,325281,203112,R
+vis3,767124,306692,219120,L,801424,285944,270713,R
+antler,476963,179977,195734,R,458685,182007,185345,L
+dimple1,340406,205422,106682,R,348777,222823,58876,L
+dimple2,459534,98028,150176,R,438786,104989,138065,L
+dimple3,437621,304754,157541,R,423415,308005,144728,L
+dimple4,400847,199521,93013,R,402247,214585,66168,L
+LH,324187,134702,182934,R,312855,146563,152677,L
+MB1,364791,154893,102163,R,352928,172890,65884,L
+MB2,454182,85278,90487,R,446883,93341,76730,L
+peak,415196,275831,66134,R,426082,294979,53835,L
+SEZ,484171,311357,47117,R,486096,317818,37401,L
+vis1,314574,183857,235749,R,315736,189507,204076,L
+vis2,365547,325281,203112,R,365644,344243,170947,L
+vis3,244462,285944,270713,R,278762,306692,219120,L
diff --git a/flybrains/data/README.md b/flybrains/data/README.md
index 6ab3c16..426317d 100644
--- a/flybrains/data/README.md
+++ b/flybrains/data/README.md
@@ -8,3 +8,10 @@ However, they are in raw voxel coordinates.
 In contrast, X1,Y1,Z1 (which contain locations in FAFB space) are in nm.
 
 See https://github.com/saalfeldlab/elm for the original source of the landmarks.
+
+`FAFB14_mirror_landmarks.csv` contains landmarks to compensate for left/right
+asymmetries in FAFB14. Pairs of landmarks (e.g. "antlers L" and "antler R")
+were placed manually to approximately corresponding locations on the left and
+right hemisphere. To generate a mirror transform, for each landmark (e.g.
+"antlers L") the corresponding landmark on the other side (e.g. "antlers R")
+was mirrored to the same side using an affine transform. 
diff --git a/flybrains/templates.py b/flybrains/templates.py
index 03db4fd..173ac96 100644
--- a/flybrains/templates.py
+++ b/flybrains/templates.py
@@ -640,11 +640,11 @@ class _JRCVNC2018F(FlyTemplateBrain):
     The JRCVNC2018U reference VNC was constructed as for JRCVNC2018F VNC but
     pooling both male and female brains.
 
-    JRCVNC2018F.surf, was generated in Amira from a 2 micron downsampled,
+    The JRCVNC2018F mesh, was generated in Amira from a 2 micron downsampled,
     Lanczos filtered version of the respective template VNC. A surface was then
     generated with a threshold level of 2800 and then smoothed and downsampled.
 
-    JRCVNC2018U.surf, was generated in Amira from a 2 micron downsampled,
+    The JRCVNC2018U mesh, was generated in Amira from a 2 micron downsampled,
     Lanczos filtered 8 bit version of the respective template VNC. A surface was
     then generated with a threshold level of 20 and then smoothed and
     downsampled.