Ena interface

app/api/ena/route.ts

@@ -0,0 +1,240 @@
+import {
+  ENAAssembly,
+  PrimaryDataApiResult,
+} from "app/apis/catalog/brc-analytics-catalog/common/entities";
+import { NextResponse } from "next/server";
+
+const items_limit = 10000;
+
+async function fetchDataFromENA(
+  url: string,
+  redirect_arg: RequestRedirect = "manual"
+): Promise<PrimaryDataApiResult> {
+  const response: Response = await fetch(url, { redirect: redirect_arg });
+  if (response.status !== 200) {
+    if (response.status === 301) {
+      console.debug(
+        `ENA Rest API has been updated, internal API call needs update: ${url}`
+      );
+      return {
+        count: 0,
+        data: [],
+        error: `ENA Rest API has been updated!!!`,
+        status: 301,
+      };
+    }
+    const errorMessageText = await response.text();
+    let errorMessage;
+    try {
+      errorMessage = JSON.parse(errorMessageText).message;
+    } catch (e) {
+      errorMessage = `errorMessageText: ${e}`;
+    }
+    return {
+      count: 0,
+      data: [],
+      error: `from ENA, ${errorMessage}`,
+      status: response.status,
+    };
+  }
+  const result = await response.json();
+  return Promise.resolve({
+    count: "count" in result ? parseInt(result.count) : result.length,
+    data: result,
+    error: "",
+    status: response.status,
+  });
+}
+
+async function processAccessionIds(filter: string): Promise<string> {
+  // Identify any GCG/GCA accession IDS in the filter string, if they exist
+  // we need to translate them into sample IDs
+  // Will convert the following string
+  // 'accession=GCA_009859065.2 AND accession=GCF_009859065.2'
+  // into
+  // (sample_accession="SAMN09946140" OR sample_accession="SAMN0994555") OR sample_accession="NO_SAMPLE"
+  // GCA_009859065.2 generated 2 sample_accession and GCF_009859065.2 generated 0 sample_accession
+
+  function getSampleAcceionIds(filter: string): {
+    accessions: string[];
+    accessionsDict: { [key: string]: string[] };
+    accessionsUrl: string;
+  } {
+    const accessions: string[] = [];
+    const accessionsDict: { [key: string]: string[] } = {};
+    const accessionRegex = /(accession\s*=\s*)("?)(GC[FA]_\d+\.\d+)("?)/g;
+    let match;
+    let counterCounter = 0;
+    while ((match = accessionRegex.exec(filter)) !== null) {
+      counterCounter++;
+      accessions.push(match[3]);
+      if (accessionsDict[match[3]]) {
+        accessionsDict[match[3]].push(
+          `${match[1]}${match[2]}${match[3]}${match[4]}`
+        );
+      } else {
+        accessionsDict[match[3]] = [
+          `${match[1]}${match[2]}${match[3]}${match[4]}`,
+        ];
+      }
+    }
+
+    if (filter.match(/\b(accession\s*=)\s*/g)?.length !== counterCounter) {
+      throw new Error(
+        `GCF/GCA syntax error, on or multiple accession id have incorrect format, should be GCF_XXXXXXX.X or GCA_XXXXXXX.X`
+      );
+    }
+    if (accessions.length === 0) {
+      accessions.push("NO_ACCESSION");
+    }
+    const queryParams = new URLSearchParams({
+      query: accessions
+        .map((item) => `assembly_set_accession="${item}"`)
+        .join(" OR "),
+    });
+    // Construct the URL for the API request to fetch sample IDs based on accession IDs
+    return {
+      accessions: accessions,
+      accessionsDict: accessionsDict,
+      accessionsUrl: `https://www.ebi.ac.uk/ena/portal/api/search?result=assembly&fields=assembly_set_accession,sample_accession&${queryParams.toString()}&format=json`,
+    };
+  }
+
+  function updateFilterWithSampleIds(
+    filter: string,
+    accessions: string[],
+    accessionsDict: { [key: string]: string[] },
+    response: { data: []; error: string; status: number }
+  ): string {
+    // Extract sample IDs from the API response and add them to the sample_ids array
+    const sample_ids: { [key: string]: string[] } = {};
+    for (const assembly of response.data as ENAAssembly[]) {
+      if (!sample_ids[assembly.assembly_set_accession]) {
+        sample_ids[assembly.assembly_set_accession] = [
+          assembly.sample_accession,
+        ];
+      } else {
+        sample_ids[assembly.assembly_set_accession].push(
+          assembly.sample_accession
+        );
+      }
+    }
+    for (const accession of accessions) {
+      if (sample_ids[accession]) {
+        for (const accessionExpression of accessionsDict[accession]) {
+          filter = filter.replace(
+            accessionExpression,
+            "(" +
+              sample_ids[accession]
+                .map((sample_id) => `sample_accession="${sample_id}"`)
+                .join(" OR ") +
+              ")"
+          );
+        }
+      } else {
+        for (const accessionExpression of accessionsDict[accession]) {
+          filter = filter.replace(
+            accessionExpression,
+            `sample_accession="NO_SAMPLE"`
+          );
+        }
+      }
+    }
+    return filter;
+  }
+
+  if (filter.includes("accession=")) {
+    // Find accessions id and create an url and mappers to
+    // convert the accession id to sample id
+    const { accessions, accessionsDict, accessionsUrl } =
+      getSampleAcceionIds(filter);
+
+    // if (accessions.length === 1 && accessions[0] === "NO_ACCESSION") {
+    //   throw new Error(
+    //     `GCF/GCA syntax error, on or multiple accession id have incorrect format, should be GCF_XXXXXXX.X or GCA_XXXXXXX.X`
+    //   );
+    // }
+
+    // Fetch the data from the API
+    const accession_response = await fetchDataFromENA(accessionsUrl);
+    if (accession_response.status !== 200) {
+      throw new Error(
+        `ENA API error: status: ${accession_response.status}, message: ${JSON.stringify(accession_response.data)}`
+      );
+    }
+
+    // Update the filter string with the sample IDs
+    return updateFilterWithSampleIds(
+      filter,
+      accessions,
+      accessionsDict,
+      accession_response
+    );
+  }
+  return filter;
+}
+
+export async function POST(request: Request): Promise<NextResponse> {
+  let { filter } = await request.json();
+  const fields = [
+    "accession",
+    "sra_md5",
+    "base_count",
+    "study_accession",
+    "sample_accession",
+    "instrument_platform",
+    "instrument_model",
+    "library_layout",
+  ];
+  try {
+    filter = await processAccessionIds(filter);
+  } catch (error) {
+    return NextResponse.json(
+      { count: 0, data: [], error: `${error}` },
+      { status: 500 }
+    );
+  }
+
+  const runQueryParams = new URLSearchParams({
+    query: filter,
+  });
+
+  const filter_url = `${runQueryParams.toString().replace(/%3D%3D/g, "%3D")}`;
+  const count_url = `https://www.ebi.ac.uk/ena/portal/api/count?result=read_run&${filter_url}&format=json`;
+
+  const count_response = await fetchDataFromENA(count_url);
+  if (count_response.status !== 200) {
+    return NextResponse.json(
+      { count: 0, data: [], error: count_response.error },
+      {
+        status: count_response.status,
+      }
+    );
+  }
+
+  const count: number = count_response.count;
+  if (count === 0) {
+    return NextResponse.json({ count: 0, data: [] }, { status: 200 });
+  }
+  if (count > items_limit) {
+    return NextResponse.json({
+      count: 0,
+      data: [],
+      error: `To many entries return: ${count}, please add filters to reduce the number of entries.`,
+    });
+  }
+
+  const url_search = `https://www.ebi.ac.uk/ena/portal/api/search?result=read_run&${filter_url}&fields=${fields.join(",")}&limit=${items_limit}&format=json`;
+  //console.debug(`Search URL: ${url_search}`);
+  const search_response = await fetchDataFromENA(url_search);
+  if (search_response.status !== 200) {
+    return NextResponse.json(search_response.data, {
+      status: search_response.status,
+    });
+  }
+
+  return NextResponse.json(
+    { count: count, data: search_response.data },
+    { status: search_response.status }
+  );
+}

app/apis/catalog/brc-analytics-catalog/common/entities.ts

@@ -46,6 +46,38 @@ export interface EntitiesResponsePagination {
   total: number;
 }
 
+export interface ENAAssembly {
+  assembly_set_accession: string;
+  sample_accession: string;
+}
+
+export interface ENAReadRuns {
+  accession: string;
+  base_count: string;
+  sample_accession: string;
+  studies_accession: string;
+}
+
+export interface PrimaryDataApiResult {
+  count: number;
+  data: [];
+  error: string;
+  status: number;
+}
+
+export interface RunReadsFields {
+  description: string;
+  name: string;
+  type: string;
+}
+
+export interface ReadRunStatistics {
+  bases: number;
+  biosamples: number;
+  read_runs: number;
+  studies: number;
+}
+
 export interface WorkflowCategory {
   category: string;
   description: string;

app/components/Entity/components/PrimaryDataViewer/primaryDataViewer.style.ts

@@ -0,0 +1,110 @@
+import { GridPaperSection } from "@databiosphere/findable-ui/lib/components/common/Section/section.styles";
+
+import styled from "@emotion/styled";
+
+export const StyledSection = styled(GridPaperSection)`
+  flex-direction: column;
+`;
+
+export const SectionContent = styled.div`
+  flex: 1;
+  display: flex;
+  flex-direction: column;
+  align-items: center;
+  img {
+    width: 600px;
+    height: 600px;
+  }
+`;
+
+export const BlinkingDots = styled.span`
+  font-weight: bold;
+  @keyframes blink {
+    0% {
+      opacity: 1;
+    }
+    33% {
+      opacity: 0;
+    }
+    66% {
+      opacity: 0;
+    }
+    100% {
+      opacity: 1;
+    }
+  }
+  .dot:nth-child(1) {
+    animation: blink 1.5s infinite;
+  }
+  .dot:nth-child(2) {
+    animation: blink 1.5s infinite 0.5s;
+  }
+  .dot:nth-child(3) {
+    animation: blink 1.5s infinite 1s;
+  }
+`;
+
+export const ReactPageNation = styled.span`
+  /* General container styling */
+  .container {
+    max-width: 800px;
+    margin: 50px auto;
+    padding: 20px;
+    background-color: #f9f9f9;
+    border-radius: 8px;
+    box-shadow: 0 4px 8px rgba(0, 0, 0, 0.1);
+    font-family: "Arial", sans-serif;
+  }
+
+  /* Blog posts list styling */
+  .list-group-item {
+    background-color: white;
+    border: 1px solid #ddd;
+    border-radius: 8px;
+    padding: 20px;
+    margin-bottom: 15px;
+    transition: box-shadow 0.3s ease;
+  }
+
+  .list-group-item p {
+    color: #666;
+    line-height: 1.6;
+  }
+
+  /* Pagination styling */
+  .pagination {
+    display: flex;
+    justify-content: center;
+    padding: 0px;
+
+    list-style: none;
+  }
+
+  .pagination li {
+    margin: 0 1px;
+    font-size: 0.7rem;
+  }
+
+  .pagination li a {
+    display: inline-block;
+    padding: 5px 10px 5px 10px;
+    background-color: white;
+    color: #212b36;
+    border-radius: 5px;
+    text-decoration: none;
+    transition:
+      background-color 0.3s ease,
+      color 0.3s ease;
+  }
+
+  .pagination li a:hover {
+    background-color: #212b36;
+    color: white;
+  }
+
+  .pagination li.active a {
+    border-color: #212b36;
+    border-style: dashed;
+    border-width: 1px;
+  }
+`;