Add Ag cohorts analysis 20240924

_posts/2024-10-02-ag3-cohorts-v20240924.md

@@ -0,0 +1,44 @@
+---
+layout: post
+title:  "Ag3 cohorts analysis version 20240924"
+tags: data
+---
+
+A new cohorts analysis version `20240924` has been released for the
+Ag3 data resource. This is now the default cohorts analysis version
+when using the `malariagen_data` [Ag3
+API](https://malariagen.github.io/malariagen-data-python/latest/Ag3.html). This
+cohorts analysis will be available for datasets up to and including Ag3.13.
+
+Please note that the new cohorts analysis may change the values of
+sample metadata columns including `taxon`, `admin1_iso`,
+`admin1_name`, `admin2_name`, and derived columns beginning `cohorts_`
+relative to previous cohorts analysis versions.
+
+To pin this cohorts analysis when accessing data:
+
+{% highlight python %}
+import malariagen_data
+
+ag3 = malariagen_data.Ag3(
+    cohorts_analysis="20240924",
+)
+{% endhighlight %}
+
+This new version introduces some key changes: 
+
+- Samples previously labeled as `gcx1` in the `taxon` field have been renamed to `bissau`:
+    - `gcx` (`gambiae complex cryptic taxa`) labels serve as placeholders for groups outside our usual taxonomic assignment 
+    - Following [Caputo et al. (2024)](https://malariagen.github.io/vobs-updates/2024/09/10/caputo.html), the `gcx1` group has been renamed to `Bissau molecular form`
+    - 291 samples previously assigned as `gcx1`, are now labeled as `bissau`.
+    - 5 previously `unassigned` samples are also relabeled as `bissau`.
+    - Cohort names have been updated, e.g. `GM-M_gcx1_2019` is now `GM-M_biss_2019`
+
+- 36 previously `unassigned` samples have been reclassified as: 32 `melas`, 2`gambiae`, 1 `fontenillei`, 1 `arabiensis`.
+
+- A location metadata error affecting the administrative region (level 1) of 119 samples has been corrected:
+    -  `admin1_iso` updated from  `UG-E` to `KE-04` 
+    -  `admin1_name` updated from  `Eastern Region` to `Busia`
+    - Cohort names have been updated, e.g. `UG-E_arab_2013` has now been relabeled to `KE-04_arab_2013`
+
+If you need to access the previous version of the cohorts analysis, you can pin it using the code in [here](https://malariagen.github.io/vobs-updates/2024/07/24/ag3-cohorts-v20240717.html).