1 00:00:00,240 --> 00:00:04,900 Hi, and welcome to the MarinaBioLab qPCR validation workbench. 2 00:00:05,600 --> 00:00:09,640 This tool turns a raw instrument export into a finished validation report. 3 00:00:11,240 --> 00:00:15,760 In about twelve minutes, I'll walk you through one complete validation, start to 4 00:00:15,800 --> 00:00:20,500 finish. You'll see where to click, what to check, and what to do when something looks 5 00:00:20,520 --> 00:00:25,860 wrong. Everything on screen is made-up training data, and the video has chapters, 6 00:00:26,220 --> 00:00:28,260 so you can jump back to any step later. 7 00:00:29,593 --> 00:00:34,953 Let's sign in. There's no password. You use your work email instead. Type your 8 00:00:35,013 --> 00:00:39,193 address into the Work email box, then press Send sign-in link. 9 00:00:41,213 --> 00:00:45,973 Open that email on the same computer you're using now, and click the link. If 10 00:00:45,993 --> 00:00:50,373 you open it on your phone instead, the link signs in your phone, so type the 11 00:00:50,433 --> 00:00:55,673 eight digit code from the email into this page. And if no email arrives, ask your 12 00:00:55,713 --> 00:00:56,673 validation lead. 13 00:00:58,033 --> 00:01:02,773 You're on the Dashboard. Let's get our bearings. The menu on the left is arranged 14 00:01:02,813 --> 00:01:07,773 in the order you'll work. Workspace, then Set up, then Data, then 15 00:01:07,853 --> 00:01:13,173 Output, and the AI co-pilot at the end. The strip across the 16 00:01:13,233 --> 00:01:18,293 top is called Your validation workflow. It shows the five stages of every 17 00:01:18,333 --> 00:01:23,733 validation, each with a short status and a suggested next action. Whenever 18 00:01:23,773 --> 00:01:26,533 you're unsure what to do next, read that text. 19 00:01:29,467 --> 00:01:34,427 Every validation lives inside a session. Think of it as a folder. It holds the 20 00:01:34,487 --> 00:01:39,707 panels you're validating, the study types, every file you upload, and everything the 21 00:01:39,727 --> 00:01:44,827 workbench produces. One session can hold several panels, so a multiplex 22 00:01:44,867 --> 00:01:50,247 validation stays together. We're on the Sessions page. Press 23 00:01:50,467 --> 00:01:55,467 Start a validation. After a moment the status says Session ready, and the header 24 00:01:55,507 --> 00:01:57,467 at the top shows which session you're in. 25 00:02:00,413 --> 00:02:05,733 Next, the plan. This is where you say what you intend to validate, before any 26 00:02:05,793 --> 00:02:10,973 numbers come in. You're on the Plan page. Tick the panels you're 27 00:02:11,013 --> 00:02:15,533 validating, then tick the study types you're running. Then press Build guided 28 00:02:15,613 --> 00:02:20,873 plan. If your lab already has a saved plan file, press Import plan 29 00:02:20,933 --> 00:02:25,633 JSON instead. Getting this right now decides how your results are organised 30 00:02:25,733 --> 00:02:26,033 later. 31 00:02:27,380 --> 00:02:32,240 Sometimes the workbench needs a decision that only a scientist can make. Then it 32 00:02:32,280 --> 00:02:36,980 shows a decision card like this one. Read the question and the reason it's asked. 33 00:02:37,560 --> 00:02:41,840 Pick the option that matches your evidence. If none fit and there's an Other 34 00:02:41,980 --> 00:02:47,100 choice, describe your own answer. Add the short explanation it asks for, 35 00:02:47,540 --> 00:02:52,880 then press Apply answer. It's saved with your name and the time. Clear 36 00:02:52,960 --> 00:02:55,020 every open card before moving on. 37 00:02:57,967 --> 00:03:03,087 Now your data. You're on the Upload page. One important 38 00:03:03,127 --> 00:03:08,107 detail. The workbench can't read a Bio-Rad .pcrd run file 39 00:03:08,187 --> 00:03:13,367 directly. Export it first. In CFX Maestro, open your run, 40 00:03:13,727 --> 00:03:18,467 go to Export, then Export All Data Sheets, and choose CSV. 41 00:03:20,287 --> 00:03:25,227 QuantStudio exports, RDML files, and Excel workbooks work too. 42 00:03:25,967 --> 00:03:29,627 If you pick the wrong file type, the message on screen reminds you what to 43 00:03:29,687 --> 00:03:30,207 export. 44 00:03:31,587 --> 00:03:35,807 Here's the habit that saves the most time. Put the study type in the file name. 45 00:03:37,967 --> 00:03:42,687 Look at the selected file. Its name includes the word Accuracy, so the 46 00:03:42,727 --> 00:03:47,287 workbench can route those rows to the Accuracy study on its own. If the name 47 00:03:47,327 --> 00:03:51,347 doesn't say which study it belongs to, that's okay, but you'll confirm it by 48 00:03:51,387 --> 00:03:54,227 hand, one group at a time, in the next step. 49 00:03:55,547 --> 00:04:00,027 Next to the file, fill in the Reagent kit catalog and the Reagent lot number. 50 00:04:00,687 --> 00:04:04,967 They're recorded with each upload for lot tracking, along with your name, and they 51 00:04:05,127 --> 00:04:10,527 never appear in a client report. Then press Save and queue 52 00:04:10,567 --> 00:04:15,567 parser. This is the step people miss. Choosing a file doesn't read it. This 53 00:04:15,607 --> 00:04:20,687 button does. Open Jobs in the header, and when the parser says succeeded, you're 54 00:04:20,727 --> 00:04:21,967 ready for mapping review. 55 00:04:24,933 --> 00:04:29,393 You're on the Mapping review page. This step decides what your numbers mean, so 56 00:04:29,793 --> 00:04:34,913 slow down here. Press Load mapper groups. The workbench 57 00:04:35,033 --> 00:04:40,353 sorts the rows from your file into groups, one per sample. Each table row is one 58 00:04:40,393 --> 00:04:45,453 group. You can see the file, the sample name, which study it went to, where that 59 00:04:45,493 --> 00:04:50,393 routing came from, how many source rows it covers, and a reason in plain words. 60 00:04:51,720 --> 00:04:55,280 Look at the Source column. It tells you how each group was routed. 61 00:04:57,320 --> 00:05:02,760 Nomenclature means it recognised the study from text in your file. Parser guess 62 00:05:02,820 --> 00:05:07,900 means the parser suggested a study role. Unresolved means it couldn't tell, so you 63 00:05:07,940 --> 00:05:12,740 decide. Positive controls and no template controls are tracked separately. 64 00:05:13,500 --> 00:05:18,120 For each group, check the sample, its source rows, and the reason before you 65 00:05:18,160 --> 00:05:19,100 accept the routing. 66 00:05:20,440 --> 00:05:25,200 Sometimes a target name in the file isn't quite right. That's what Manual target 67 00:05:25,240 --> 00:05:30,680 correction is for. Press Load server target choices. Pick the measurement 68 00:05:30,720 --> 00:05:35,480 to fix, choose the correct target from the list, and type a short scientific reason. 69 00:05:36,240 --> 00:05:41,480 Press Preview correction to see the before and after values. If it looks right, tick 70 00:05:41,520 --> 00:05:46,800 the confirmation box and press Confirm target correction. Every correction and 71 00:05:46,840 --> 00:05:48,560 its reason stays in the history. 72 00:05:51,513 --> 00:05:55,753 Now let's calculate the results. Once you confirm the mapping and the session is 73 00:05:55,813 --> 00:05:59,373 ready, the workbench usually queues the calculation for you. 74 00:06:01,233 --> 00:06:06,413 To check, open Jobs from the header. If nothing has started, go to the Dashboard 75 00:06:06,473 --> 00:06:11,653 and press Queue next output job. If something is blocking it, the message says 76 00:06:11,713 --> 00:06:16,413 exactly what's missing. You don't have to wait here. The Results page shows the 77 00:06:16,473 --> 00:06:17,713 saved run when it's ready. 78 00:06:19,067 --> 00:06:24,307 You're on the Results page. Across the top are the headline numbers. Overall status, 79 00:06:24,687 --> 00:06:29,887 mean detection rate, mean R squared, the largest Cq gap, and how many 80 00:06:29,947 --> 00:06:35,387 targets and rows were used. Be careful with one of them. Mean 81 00:06:35,427 --> 00:06:39,847 detection rate averages the rates from each study summary row. It's not one 82 00:06:39,887 --> 00:06:44,947 percentage from every well. Treat these as a summary, not a verdict. The real 83 00:06:45,107 --> 00:06:47,007 story is in the study tables below. 84 00:06:48,313 --> 00:06:53,533 Results has a tab for each study type. Each table has one row per metric. 85 00:06:54,053 --> 00:06:58,813 You'll see the study, the metric, the observed value, the criterion it's judged 86 00:06:58,873 --> 00:07:04,013 against, the status, and the source rows behind it. Cq 87 00:07:04,073 --> 00:07:09,253 values show two decimal places. If a study was planned but has no data yet, 88 00:07:09,653 --> 00:07:15,133 its row says Needs review, Blocked, or N A instead of a number. 89 00:07:15,913 --> 00:07:18,853 The workbench never fills in a number it can't back up. 90 00:07:20,167 --> 00:07:23,987 Want the evidence behind a result? Open the criteria review table. 91 00:07:25,707 --> 00:07:30,567 Each row names the study, the metric, the observed value, the criterion, 92 00:07:31,027 --> 00:07:36,286 and the status. The Detail column explains how it was judged. The Source 93 00:07:36,387 --> 00:07:40,847 rows column points back to the measurements it came from. Scroll sideways 94 00:07:40,887 --> 00:07:46,047 if the table is wide. Before you sign off anything marked for review, follow those 95 00:07:46,087 --> 00:07:48,187 references back to the saved measurements. 96 00:07:49,533 --> 00:07:54,233 Let's trace one number back to the raw data. Go to the Dashboard and find the 97 00:07:54,253 --> 00:07:59,413 Workbook explorer. Pick a sheet, then pick a row, and its 98 00:07:59,473 --> 00:08:04,973 details open. Every cell shows the formula that produced it and the source rows it 99 00:08:05,113 --> 00:08:09,813 used. Source row comparison puts the calculated value next to the original 100 00:08:09,893 --> 00:08:15,353 measurements. Explain formula describes how the detection rate is worked out. It's 101 00:08:15,393 --> 00:08:18,833 a reminder, not a substitute for checking the sources yourself. 102 00:08:20,167 --> 00:08:25,467 The figures come from the same result run. Target detection rate, mean Cq by 103 00:08:25,527 --> 00:08:30,547 target, and the standard curves. Where a panel reports a semi quantitative 104 00:08:30,627 --> 00:08:34,807 range, the bands come from the fitted curve for that target rather than a fixed 105 00:08:34,927 --> 00:08:39,907 table, and a fit too weak to separate the bands is reported qualitatively instead. 106 00:08:40,747 --> 00:08:44,767 If a figure looks wrong, the repair is upstream in the data or the mapping. 107 00:08:47,680 --> 00:08:52,180 Now let's check the plate. You're on the Plate map page. It's built from the 108 00:08:52,240 --> 00:08:56,100 calculated results, so it appears after the calculation has run. 109 00:08:57,840 --> 00:09:02,380 Use it to confirm your controls and replicates landed where you expected. Each 110 00:09:02,440 --> 00:09:07,640 well shows its call. Click any well to see the sample, the target, the Cq 111 00:09:07,700 --> 00:09:12,940 value, and the source rows behind it. If a sample ended up in the wrong well, you'll 112 00:09:13,020 --> 00:09:13,660 spot it here. 113 00:09:16,567 --> 00:09:21,727 Now the part only a scientist can do. Accepting the result. You'll find 114 00:09:21,767 --> 00:09:25,327 Result-run review on the Dashboard and on the Results page. 115 00:09:27,367 --> 00:09:31,767 Once you've checked the evidence, pick a rationale code, write your scientific 116 00:09:31,847 --> 00:09:37,247 rationale, and press Approve result. That puts your name against this exact 117 00:09:37,287 --> 00:09:42,647 result package. For a known deviation, record an Approved deviation, or 118 00:09:42,807 --> 00:09:47,827 file an exception with the target, the study, your summary, and the source rows. 119 00:09:49,667 --> 00:09:53,867 Accepting a result is not the same as approving a report. That comes later. 120 00:09:56,780 --> 00:10:01,160 The report is built from the results you just accepted. Before you generate it, 121 00:10:01,540 --> 00:10:06,800 choose who it's for. Scientist review gives the full detail, 122 00:10:06,900 --> 00:10:12,220 with every table and source reference. Client final keeps the study conclusions 123 00:10:12,240 --> 00:10:17,420 and the key performance results, including precision standard deviations and CVs, and 124 00:10:17,520 --> 00:10:22,780 trims internal detail. Both use exactly the same calculations. Then 125 00:10:22,820 --> 00:10:27,800 choose Marina brand or White label. Choosing Client final 126 00:10:27,900 --> 00:10:30,980 describes the audience. It doesn't approve the report. 127 00:10:32,353 --> 00:10:36,493 You're on the Report page. There are two kinds of fix here, and it matters which 128 00:10:36,573 --> 00:10:41,773 one you use. If the wording of a section is wrong, press Edit 129 00:10:41,833 --> 00:10:46,813 section, change it, and press Save section. That's what a draft is for. 130 00:10:47,653 --> 00:10:51,593 The Linked to this section panel keeps its tables and figures attached. 131 00:10:53,153 --> 00:10:58,173 If a number or a table is wrong, don't edit the report. Fix the data or the 132 00:10:58,233 --> 00:11:02,833 mapping upstream, generate the results again, and the report rebuilds itself. 133 00:11:04,167 --> 00:11:09,587 This strip shows where the report is in its life. Preview, QA, independent 134 00:11:09,627 --> 00:11:14,947 review, then Approved. Right now it's a Draft. Every 135 00:11:15,067 --> 00:11:20,327 draft download has a DRAFT watermark on each page, so please don't send drafts to 136 00:11:20,367 --> 00:11:25,827 clients. To become Approved, the file passes its integrity check, 137 00:11:26,207 --> 00:11:30,847 the governing policy is ratified, the release checks pass, and an independent 138 00:11:30,887 --> 00:11:36,387 reviewer approves that exact version. Until then, Approve certified report 139 00:11:36,506 --> 00:11:41,587 stays disabled, and the list underneath says what's missing. The automatic QA 140 00:11:41,707 --> 00:11:45,967 pass only gives advice. It can't approve or reject anything. 141 00:11:48,920 --> 00:11:54,200 You're on the Downloads page, where every finished file lives. Each file shows its 142 00:11:54,260 --> 00:11:58,760 type, the result it came from, and a short fingerprint so you can tell versions 143 00:11:58,800 --> 00:12:03,460 apart. Before you send anything, read the button. 144 00:12:04,260 --> 00:12:09,440 Internal preview means on screen only. Awaiting approval means the report isn't 145 00:12:09,480 --> 00:12:14,900 approved yet. On the Report page, the same file reads Download DRAFT 146 00:12:15,000 --> 00:12:20,460 watermarked PDF until approval, and Download certified report PDF 147 00:12:20,640 --> 00:12:24,020 after it. If it says draft, it's not for a client. 148 00:12:26,946 --> 00:12:31,606 Let's look at the Panel registry. It's the reference library the workbench uses. 149 00:12:33,226 --> 00:12:38,486 Pick a panel and you'll see its targets, each with its channel or dye, its oligo 150 00:12:38,586 --> 00:12:43,646 mix, and its LoD floor. The panel row shows how many study types it 151 00:12:43,686 --> 00:12:49,026 supports and its default Cq cutoff. The key word is default. 152 00:12:49,646 --> 00:12:53,766 Whether a well counts as detected also depends on the amplification status and 153 00:12:53,806 --> 00:12:58,106 the run settings. Every session records the registry version it used. 154 00:13:01,053 --> 00:13:05,473 Now the AI co-pilot. Open it from the header while your session is selected. 155 00:13:07,313 --> 00:13:12,353 A good first question is, what evidence do we have for this session, or which review 156 00:13:12,413 --> 00:13:17,693 steps are still open. Wait for the full reply. Then, and this matters, 157 00:13:18,173 --> 00:13:22,873 check its references and the values it quotes against the tables. A reply about 158 00:13:22,913 --> 00:13:27,813 the whole session doesn't tell you the result for one target or study. Ask about 159 00:13:27,853 --> 00:13:29,073 the specific thing you need. 160 00:13:30,366 --> 00:13:35,146 Treat every co-pilot answer as a draft for you to review, not a conclusion. 161 00:13:35,826 --> 00:13:41,226 Compare what it cites with the saved tables and source rows. If it 162 00:13:41,246 --> 00:13:46,666 doesn't answer your question, ask a narrower one, or use Discuss this value on 163 00:13:46,706 --> 00:13:51,386 the exact result cell. And remember, the co-pilot can't fill in missing 164 00:13:51,426 --> 00:13:56,426 measurements, make study decisions, or approve anything. Those still happen in 165 00:13:56,466 --> 00:13:58,226 the validation record, by you. 166 00:14:01,146 --> 00:14:06,406 What if you get stuck? First, read the written status on the step you're on. It 167 00:14:06,466 --> 00:14:11,966 names the blocker and suggests the next action. If something is taking a 168 00:14:12,006 --> 00:14:16,366 long time, open Jobs in the header to see what's running and whether anything 169 00:14:16,506 --> 00:14:21,946 failed. And if a number looks wrong, send your validation lead three things. The 170 00:14:21,986 --> 00:14:25,986 session name from the header, the file name, and what you expected to see 171 00:14:26,046 --> 00:14:26,606 instead. 172 00:14:27,993 --> 00:14:32,813 And that's the whole loop. Start a validation. Set your plan and answer the 173 00:14:32,833 --> 00:14:37,533 decision cards. Upload your file with the study type in the name. Review the 174 00:14:37,573 --> 00:14:42,953 mapping. Generate and approve the results. Check the plate map. Draft the report, 175 00:14:43,313 --> 00:14:48,453 get it reviewed, and download the approved copy. Come back to any 176 00:14:48,513 --> 00:14:52,633 chapter whenever you need it. Thanks for watching, and if anything looks wrong, 177 00:14:52,973 --> 00:14:54,013 please send it through.