From 5eac092c92b3f0d44a49e2b3cbc340540c14a7a5 Mon Sep 17 00:00:00 2001
From: ivan-aksamentov <ivan.aksamentov@gmail.com>
Date: Thu, 7 Dec 2023 11:22:26 +0100
Subject: [PATCH 1/2] feat(web): sort dataset suggestions by mean score across
 sequences

---
 .../src/hooks/useRunSeqAutodetect.ts          | 36 +++++++++++++++++--
 .../src/state/autodetect.state.ts             | 13 +------
 2 files changed, 34 insertions(+), 15 deletions(-)

diff --git a/packages_rs/nextclade-web/src/hooks/useRunSeqAutodetect.ts b/packages_rs/nextclade-web/src/hooks/useRunSeqAutodetect.ts
index 24aceb5c6..23a4b0db2 100644
--- a/packages_rs/nextclade-web/src/hooks/useRunSeqAutodetect.ts
+++ b/packages_rs/nextclade-web/src/hooks/useRunSeqAutodetect.ts
@@ -1,4 +1,4 @@
-import { first, get, isNil, sortBy } from 'lodash'
+import { first, get, isNil, mean, sortBy, uniq } from 'lodash'
 import type { Subscription } from 'observable-fns'
 import { useMemo } from 'react'
 import { useRecoilCallback, useRecoilValue } from 'recoil'
@@ -9,7 +9,6 @@ import {
   autodetectResultsAtom,
   AutodetectRunState,
   autodetectRunStateAtom,
-  groupByDatasets,
   minimizerIndexAtom,
 } from 'src/state/autodetect.state'
 import { datasetsAtom, minimizerIndexVersionAtom } from 'src/state/dataset.state'
@@ -114,6 +113,32 @@ export class SeqAutodetectWasmWorker {
   }
 }
 
+export function groupByDatasets(records: MinimizerSearchRecord[]) {
+  const names = uniq(records.flatMap((record) => record.result.datasets.map((dataset) => dataset.name)))
+  let byDataset: Record<string, { records: MinimizerSearchRecord[]; meanScore: number }> = {}
+  // eslint-disable-next-line no-loops/no-loops
+  for (const name of names) {
+    // Find sequence records which match this dataset
+    const selectedRecords = records.filter((record) => record.result.datasets.some((dataset) => dataset.name === name))
+
+    // Get scores for sequence records which match this dataset
+    const scores = selectedRecords.map((record) => {
+      const dataset = record.result.datasets.find((ds) => ds.name === name)
+      return dataset?.score ?? 0
+    })
+    const meanScore = mean(scores)
+
+    byDataset = { ...byDataset, [name]: { records: selectedRecords, meanScore } }
+  }
+
+  console.info(
+    sortBy(Object.entries(byDataset), ([_, val]) => -val.meanScore)
+      .map(([name, val]) => `${name.padEnd(60)} ${val.meanScore.toFixed(4)}`)
+      .join('\n'),
+  )
+  return byDataset
+}
+
 export function useDatasetSuggestionResults() {
   const { datasets } = useRecoilValue(datasetsAtom)
   const autodetectResults = useRecoilValue(autodetectResultsAtom)
@@ -128,7 +153,12 @@ export function useDatasetSuggestionResults() {
       Object.entries(recordsByDataset).some(([dataset, _]) => dataset === candidate.path),
     )
 
-    itemsInclude = sortBy(itemsInclude, (dataset) => -get(recordsByDataset, dataset.path, []).length)
+    itemsInclude = sortBy(itemsInclude, (dataset) => {
+      const record = get(recordsByDataset, dataset.path)
+      return -record.meanScore ?? 0
+    })
+
+    itemsInclude = sortBy(itemsInclude, (dataset) => -(get(recordsByDataset, dataset.path)?.records?.length ?? 0))
 
     const itemsNotInclude = datasets.filter((candidate) => !itemsInclude.map((it) => it.path).includes(candidate.path))
 
diff --git a/packages_rs/nextclade-web/src/state/autodetect.state.ts b/packages_rs/nextclade-web/src/state/autodetect.state.ts
index 127d0932a..5f8e2bd5e 100644
--- a/packages_rs/nextclade-web/src/state/autodetect.state.ts
+++ b/packages_rs/nextclade-web/src/state/autodetect.state.ts
@@ -1,5 +1,4 @@
-/* eslint-disable no-loops/no-loops */
-import { isEmpty, isNil, uniq } from 'lodash'
+import { isEmpty, isNil } from 'lodash'
 import { atom, atomFamily, DefaultValue, selector, selectorFamily } from 'recoil'
 import { isDefaultValue } from 'src/state/utils/isDefaultValue'
 import type { MinimizerIndexJson, MinimizerSearchRecord } from 'src/types'
@@ -49,16 +48,6 @@ export const autodetectResultByIndexAtom = selectorFamily<MinimizerSearchRecord,
 // Dataset ID to use for when dataset is not autodetected
 export const DATASET_ID_UNDETECTED = 'undetected'
 
-export function groupByDatasets(records: MinimizerSearchRecord[]): Record<string, MinimizerSearchRecord[]> {
-  const names = uniq(records.flatMap((record) => record.result.datasets.map((dataset) => dataset.name)))
-  let byDataset = {}
-  for (const name of names) {
-    const selectedRecords = records.filter((record) => record.result.datasets.some((dataset) => dataset.name === name))
-    byDataset = { ...byDataset, [name]: selectedRecords }
-  }
-  return byDataset
-}
-
 // Select autodetect results by dataset name
 export const autodetectResultsByDatasetAtom = selectorFamily<MinimizerSearchRecord[] | undefined, string>({
   key: 'autodetectResultByDatasetAtom',

From e2d713ca62dae37422ad8d5ab70ffd69f90643e8 Mon Sep 17 00:00:00 2001
From: ivan-aksamentov <ivan.aksamentov@gmail.com>
Date: Thu, 7 Dec 2023 11:54:48 +0100
Subject: [PATCH 2/2] refactor: lint

---
 packages_rs/nextclade-web/src/hooks/useRunSeqAutodetect.ts | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/packages_rs/nextclade-web/src/hooks/useRunSeqAutodetect.ts b/packages_rs/nextclade-web/src/hooks/useRunSeqAutodetect.ts
index 23a4b0db2..b6d642d6e 100644
--- a/packages_rs/nextclade-web/src/hooks/useRunSeqAutodetect.ts
+++ b/packages_rs/nextclade-web/src/hooks/useRunSeqAutodetect.ts
@@ -155,7 +155,7 @@ export function useDatasetSuggestionResults() {
 
     itemsInclude = sortBy(itemsInclude, (dataset) => {
       const record = get(recordsByDataset, dataset.path)
-      return -record.meanScore ?? 0
+      return -record.meanScore
     })
 
     itemsInclude = sortBy(itemsInclude, (dataset) => -(get(recordsByDataset, dataset.path)?.records?.length ?? 0))