Applied AI-Powered Bioinformatics Tools

Core Service Intake Validator

Last reviewed

Beginner

What you'll learn

~40 min
  • Ask an AI tool, in plan mode, for a small program that checks service requests (a lab's form and its sample sheet) against your facility's written service requirements and lists what to follow up before the samples arrive
  • Read its plan for the rules that matter at intake: a blank is missing, a value at a minimum meets it, units are flagged and never converted, and a safety answer that needs a person is listed for one, never ruled on
  • Check the list against the practice requests, request by request and sample by sample
  • See why a flag, or the lack of one, is not a decision, and keep the decision on each request for a person

The story, and what you will end up with

It is Monday 16 November 2026, before 8:30, at the intake desk of a sequencing and proteomics core facility. Labs request a service online before they bring their samples: one form per request, with the lab, the PI, a contact, the account to charge and its approver, the service, how many samples, and two safety questions, plus a sample sheet with one row for each tube. Five requests came in on Thursday and Friday, and the labs plan to drop their samples off this week, so anything you need from them has to go out this morning.

You have already done what comes first. The export, the spreadsheet your request system saves of the week’s requests, holds every request from Thursday and Friday, none withdrawn or sent twice; nothing in the inbox corrects any of them; no tubes have arrived ahead of their requests; and the service requirements you will check against are the current ones, in effect since 1 September 2026. You also know which labs have safety registrations and approvals on file with the facility; the request files do not say, so the list cannot show them. Until now you checked each request by eye against those requirements.

By the end of this lesson you will have a small program on your own computer, the intake check, that you run each time requests come in. It reads the requests and your facility’s requirements and writes a list you open in your browser and print: for each request, what differs from the requirements and what to follow up, with a blank place for your decision. It decides nothing and sends nothing. You build it by asking your AI tool in plain English, and you check it against this lesson’s numbers.

🔍Words used in this lessonclick to expand
  • Request: one lab’s form for one service, with its sample sheet. Each has a code such as R-1101.
  • PI: principal investigator, the head of the lab that sends the request.
  • Human-derived material, hazardous material: the form’s two safety questions, which the lab answers Yes or No for its samples.
  • Service requirements: what the facility asks of every sample for a service, written down by whoever owns those rules at your facility. In this lesson they are a minimum volume and a minimum total amount for each service.
  • Total amount: volume times concentration. 25 µL at 18 ng/µL is 450 ng.
  • µL, ng/µL, µg/µL: microlitres; nanograms per microlitre; micrograms per microlitre. The practice files write them uL, ng/uL and ug/uL. 2 mg/mL is the same concentration as 2 µg/µL, but it is not the unit the form asks for.
  • Account approver: the person named on the form as approving the charge. A name in that field is not an approval; whether the account is open and the charge allowed is checked with whoever handles accounts at your facility.
  • Cleared for drop-off, ask the lab, hold: the three choices this lesson leaves for you on each request. Cleared for drop-off means the lab may bring the tubes. It is not acceptance: where your facility checks the tubes when they arrive, acceptance comes after that check. Which of these you may decide yourself is set by your facility: a safety answer is followed up as your procedures say, by whoever they name, and an account is checked by whoever handles accounts.

Before you start

First, if your AI tool is still open, leave it: type /exit and press Enter (if you use Antigravity, press Ctrl+D with nothing typed instead), and wait until you see the $ or % prompt. The next line goes there, in the terminal, not in the AI tool. It makes a folder for this lesson beside your practice folder and goes into it:

Terminal window
mkdir -p ~/bio-workbench/07-intake-check && cd ~/bio-workbench/07-intake-check

Start your AI tool with the start line you kept from the setup lesson, the one with approval prompts off. Then ask it to set the folder up:

Set this folder up so we can undo any change you make, download https://learn.intellimetrics.net/starters/biotech/07/intake-requests.zip and unpack it into a data folder, and save a checkpoint.

Look for it saying three things: the folder is set up with Git (the program that saves snapshots, from setup), three files are in the data folder (requests.csv, samples.csv and service-requirements.csv), and a first snapshot, your checkpoint, is saved. Later you can ask whether the files have changed since then. If it does not mention one of them, ask: “Did you save a checkpoint with the files in it?” It may also keep the zip file in the data folder, which is fine; if it says it kept one somewhere outside this folder, ask it to delete that copy. If the download does not work, download intake-requests.zip in your browser, check where it went (the browser’s list of downloads shows the folder), and tell your AI tool, for example: “intake-requests.zip is in my Downloads folder. Unpack that one file into a data folder here, and save a checkpoint.” In Windows PowerShell or a desktop app, “my Downloads folder” works as is; only say “my Windows Downloads folder” if you use Ubuntu (WSL) on Windows, which keeps its own Downloads folder apart from Windows.

The requirements are this practice facility’s own rules, not anyone’s standard; yours are in your facility’s own documents. The practice facility has one more rule, about safety answers: a request with a safety answer listed for follow-up is not to be cleared for drop-off until the safety follow-up line is signed. The intake check you build prints that rule as a line beside each such request; it still decides nothing and marks no hold.

The box below says what is in the practice files; later steps refer to them.

ℹWhat is in intake-requests.zipclick to expand

Three files, each of which opens in a spreadsheet program.

service-requirements.csv, this practice facility’s made-up rules, in effect from 1 September 2026:

CodeServiceSample typeMinimum volumeMinimum total amountConcentration given in
GDNA-SEQWhole-genome sequencingGenomic DNA20 µL500 ngng/µL
RNA-SEQRNA sequencingTotal RNA15 µL250 ngng/µL
PROT-IDProtein identificationProtein in solution10 µL5 µgµg/µL

requests.csv, one row per request, and samples.csv, one row per tube, 26 rows in all:

RequestLabServiceDeclaredRows on the sheetWhat the form says
R-1101Marsh labGDNA-SEQ66: MRS-01 to MRS-06Hazardous material No, but the hazard details say “Contains azide”
R-1102Ivanova labRNA-SEQ88: IVA-R1, IVA-R2 twice, IVA-R4 to IVA-R8Account approver blank; human-derived material Yes
R-1103Teo labPROT-ID45: TEO-1 to TEO-5Account blank; human-derived material left blank
R-1104Quill labSEQ-WGS33: QL-1 to QL-3Hazardous material Yes, no details
R-1105Brandt labRNA-SEQ44: BR-1 to BR-4Complete

Every sample row gives a volume, a concentration and its unit, as the lab reported them. A few of them, worked out:

SampleVolumeConcentrationTotal amount
MRS-0325 µL18 ng/µL450 ng
MRS-0420 µL25 ng/µL500 ng
MRS-0512 µL60 ng/µL720 ng
MRS-0630 µL“~50” ng/µLnone: not a number
IVA-R415 µL100 ng/µL1,500 ng
IVA-R615 µL12 ng/µL180 ng
IVA-R8blank60 ng/µLnone: no volume
TEO-210 µL0.5 µg/µL5 µg
TEO-420 µL2 mg/mLnone: not the service’s unit
TEO-512 µL0.3 µg/µL3.6 µg

Everything in these files is made up for this course: the facility, its services and every number in its requirements, the labs, people, email addresses, accounts, sample names and measurements. Each lesson in this track has its own practice files, with their own people and dates.

Switch to plan mode

Your AI tool is already open. Switch to plan mode now: Shift+Tab until it says plan, or /plan (the keys for each tool). In plan mode it usually reads your files, asks its questions and shows you a plan before it builds anything; the order can vary.

The conversation

Each message below is an example. Say it your own way.

Say what you want

Say what you do, which files to use, and what you want from them:

I take in submissions at our sequencing and proteomics core. Build a small program that checks the requests in data against our service requirements and lists what to follow up before samples arrive. Ask me what you need first.

Your AI tool usually reads the three files and comes back with questions. It may also list, request by request, what it has already noticed.

Answer its questions

Answer each in a sentence. These are questions the tools often ask, with one way to answer each. Yours may ask them in a different order or show a list of choices:

If it asksYou might say
Does a sample exactly at a minimum pass?Yes, a value exactly at a minimum meets it.
Is SEQ-WGS another name for GDNA-SEQ?Flag it so I can ask the lab; never guess the code.
What happens when human-derived or hazardous material is Yes?List a Yes for follow-up under our procedures; it isn’t a fault.
What if a safety answer is blank?Show it as unanswered, beside the other safety answers. Don’t mark holds; our procedures decide.
Should a missing account or approver be followed up?Yes, a missing account or approver is a follow-up.
Convert 2 mg/mL to µg/µL?Flag it so I can ask the lab; don’t convert it.
Are there other substances it should always flag?No substance list. Show any hazard details beside a No, and a Yes with none.
Any rules that are not in the file?No other requirements: only the file’s, plus the answers I’ve given you.
Who reads it, and in what form?Just me. A page I can open in my browser and print, one section per request.
Run once, or each time?I’ll run it each time requests come in.

If it asks several questions at once, answer the ones about what counts as missing, safety and units from your facility’s own procedures, a sentence each, and accept its suggestions only about how the page looks. If you cannot tell which kind a question is, treat it as a rule. If it asks something not listed here, answer it the same way, in a sentence.

When you have answered, give it the practice facility’s rule about safety answers, in a message of its own; the files do not hold it:

Beside every request with a safety answer to follow up, print this line: Not to be cleared for drop-off until the safety follow-up line is signed.

Read the plan

Next it usually shows you a plan: what it will build and how it will check it. Read it for six things:

  1. Every request and every sample row is checked against the requirements for the service it names, including the count of rows against the number declared, and sample names that appear twice.
  2. A blank is shown as missing, never as zero or as No; a blank safety answer is shown as unanswered.
  3. A value exactly at a minimum meets it.
  4. A unit that differs from the service’s, or a value that is not a number, is flagged and not converted or guessed; a service code that is not on the requirements list is flagged, and its samples are not checked against another service’s numbers.
  5. Every safety answer that needs a person is listed apart, for follow-up under your procedures: a Yes, a blank, a Yes with no details, or details beside a No. None is called a fault, and no request is marked for a hold. Beside each request with one, the list prints the line you gave it: not to be cleared for drop-off until the safety follow-up line is signed.
  6. No request is called ready, accepted or on hold, each has a blank place for your decision, and your three files are read and left as they are.

A first plan often leaves one of these out, most often item 6: it may sort the requests into groups such as “ready” and “follow up”, or mark one for a hold. Say so in a sentence, in your own words, for example:

Don't call any request ready. Leave a blank place on each request for my decision: cleared for drop-off, ask the lab, or hold, with my name and date.

When the plan has all six, approve it: if your AI tool offers a choice to go ahead, pick it; otherwise say “Yes, go ahead and build it.” In Codex, press Shift+Tab first to leave plan mode. In Copilot, saying it brings up a choice to go ahead; pick that. Let it build. It may run what it built and tell you how to run it yourself; either way, the next step asks it to. When it has finished, ask it: “Did you install anything, and where?” Nothing, or something inside this lesson’s folder, is fine for these lessons. If it names anywhere else on your computer, such as a copy of the download or a plan file of its own, ask it what that is for, and to remove it if you do not want it there.

Look at what it built

Ask it to run what it built:

Run it and show me the list.

It usually shows the list in the terminal and says where the page is saved. To see the page itself, ask: “Open the list in my browser.” If it gives you a file location or an address without opening it, copy that into your browser’s address bar and press Enter. If it starts the page at a local address, one ending in a number, and says the address is already in use, say: “Something else is using that address. Don’t stop it. Use a different address for my page.” If your AI tool does not take a message after opening the page, wait a minute. If it still does not, close the terminal window, open a new one, type this lesson’s folder line again, start your tool, and say: “Start the page again so it keeps running while we talk.” The page and your saved work are files in the folder, so nothing is lost. When you have finished with the page, ask: “Stop the list page if it is still running.”

Your list will be worded and laid out in its own way. Check it for these:

  • There are five requests, R-1101 to R-1105, and every one has a blank place for your decision, with nothing ticked.
  • R-1101 lists MRS-03 at 450 ng, below the 500 ng minimum; MRS-05 at 12 µL, below the 20 µL minimum; MRS-06’s concentration, “~50”, as not a number, with no total worked out; and, with the safety answers, the hazard answer No beside details that say “Contains azide”. MRS-04, at exactly 20 µL and 500 ng, is not flagged.
  • R-1102 lists the missing account approver, IVA-R2 appearing twice, IVA-R6 at 180 ng below the 250 ng minimum, and IVA-R8’s volume as missing, not as 0 and not as below the minimum. Its Yes to human-derived material is listed with the safety answers, for follow-up under your procedures. IVA-R4, at exactly 15 µL, is not flagged.
  • R-1103 shows the human-derived question as unanswered, with the safety answers, and is not marked for a hold. It also lists the missing account, 5 sample rows against 4 declared, TEO-4’s mg/mL as not the service’s unit, with no total worked out, and TEO-5 at 3.6 µg below the 5 µg minimum. TEO-2, at exactly 5 µg, is not flagged.
  • R-1104 lists SEQ-WGS as not one of the facility’s service codes, with its samples not checked, and, with the safety answers, its Yes to hazardous material with no details given.
  • R-1105 has nothing to follow up, and still has a blank place for your decision.
  • R-1101, R-1102, R-1103 and R-1104, the four with a safety answer listed, each carry the line not to be cleared for drop-off until the safety follow-up line is signed. R-1105 does not, and no request is marked for a hold.
  • When you ask “Have my request files changed since the checkpoint?”, it compares them with the checkpoint and says no.

It may also tell you that IVA-R2 leaves 7 different sample names against 8 declared. Whether that is one tube entered twice or two tubes with the same name is a question for the lab, not something the list can tell. And once the Quill lab says which service R-1104 is for, run the check again on the corrected request, so its samples are checked against that service’s numbers.

If something on the list is not true, tell your AI tool in a sentence what you see and what you expected, for example: “TEO-4 shows 40 µg. Don’t convert mg/mL; flag it so I can ask the lab.” If it marks a request for a hold, say: “Don’t mark any request for a hold; a person decides that. Show a blank safety answer as unanswered, beside the other safety answers.”

Give each sign-off its own line

Each request has one place for a decision. At your facility, following up a safety answer and checking an account may be other people’s jobs, each signed for separately. Ask for a line for each:

Add two lines to each request's decision place, each with a name and date: safety answers followed up by, and account checked by. Then show me R-1105.

Check that every request now has three lines with a name and date each (the decision, the safety follow-up and the account check), all blank, and that R-1105 has them too, though it has nothing to follow up. Its account has not been checked yet either; the list only says the form was filled in.

Do not tick cleared for drop-off on a request that carries the safety line until the person your procedures name has followed the safety answer up and signed the safety follow-up line. Until then, its decision is ask the lab or hold, and which of those is set by your procedures.

Notice what a flag is

MRS-03 is flagged: 450 ng against a minimum of 500. Now try this:

Make a copy of data/service-requirements.csv in this folder. In the copy only, change the GDNA-SEQ minimum total amount to 400. Check the requests against the copy, and tell me what changes.

Before you read on: what do you expect to change, and what about the tube itself?

One flag goes away: MRS-03 now meets the minimum. Nothing else changes. R-1101 still has its hazard answer, MRS-05’s volume and MRS-06’s “~50” to ask about, and SEQ-WGS is still not checked. Nothing about MRS-03 changed. It is the same 25 µL at 18 ng/µL, as the lab reported it. Only the written rule moved.

The program compares a form with whatever the requirements file says, so its flags are only as current as that file. Facility requirements are changed by whoever owns them, with a date they take effect, for every request after that date, never to clear one request’s flag. If you and your manager decide to let the lab bring MRS-03 as it is, that decision, who made it and when, is written on R-1101, in the place for your decision. The requirements stay as they are.

The other direction matters as much. R-1105 has nothing on the list, but that says nothing about the tubes or whether its account is open, and every number on its form is the lab’s own.

A request with nothing on the list has matched the facility’s written requirements on paper, and no more; clearing it for drop-off, asking the lab or holding it is a person’s decision.

When you have looked, remove the copy: “Delete that copy, and anything you made outside this folder for it; I’m done with it.”

One real run

Here is part of what one AI tool said after the experiment above, with Claude Code on 11 November 2026. Yours will word it differently; check your list against the checks above, not against this.

With the GDNA-SEQ minimum at 400 ng, one finding goes away:
- **R-1101, Marsh lab:** MRS-03 now passes. Its total of 450 ng meets
the 400 ng minimum, where before it fell short of 500.

The labs and samples it talks about are made up for this course.

“Passes” here, like “meets requirements” on your list, means only that the number on the form is at least the number in the file; nothing about the sample changed.

How real intake differs

The practice files are simpler than any real submission:

  • Check whether your facility’s requirements ask for more than a volume and a total amount, such as quality figures, a buffer the facility can work with, or a stated way of measuring the concentration. The practice requirements leave those out. To have the program check one of yours, add it to your copy of the requirements as your facility’s own requirement, and ask your AI tool, while only the practice files are in the folder, to check that column too.
  • Every number on a form is what the lab reported. Check whether your facility measures again when the tubes arrive; if it does, that check, not this list, is where a sample is accepted. A value exactly at a minimum meets the written rule; whether to trust a lab’s reading that sits right on it is your judgement.
  • The account and approver on the form are the lab’s entries. Your procedures say who checks them, and when; the list does not.
  • The practice form asks two Yes or No questions about safety; check what your facility’s own form asks. What a Yes, a blank or a contradicted No leads to, what must be in place before the tubes come in (check what your procedures require for a human-derived Yes), and who decides, is set by your facility’s own procedures and the people they name. The list carries the answer; it never rules on it. Check also what yours say a blank or contradicted safety answer does: an automatic hold, cleared only by the person they name, or a decision the desk makes. Their rule replaces the program’s line: ask your AI tool to change it while only the practice files are in the folder.
  • Real sample names go wrong in ways these do not. A spreadsheet can turn a name into a date or drop its leading zeros; two names can differ only in capitals or a trailing space; a name can repeat one from the same lab’s earlier request. A name can also hold something that identifies a person, such as initials or a date of birth, which is a question for your facility’s procedures before it is a naming problem.
  • Look at how values arrive in your own requests, for example as “N/A”, “TBD”, “see email”, a range, or a total instead of a concentration. The list flags them and never guesses.
  • Declining a request is not one of the three choices on this list; your procedures say who may decline one.

Keep going

Small changes do not need plan mode. Say each in a sentence:

  1. “Draft an email to the Teo lab asking for what’s missing on R-1103. Say nothing about whether we’ll take the samples, prices or turnaround.” Read the draft before anything goes out. It should ask for the human-derived answer, the account, which sample count is right, TEO-4 in µg/µL, and whether more of TEO-5 is available, and it should promise nothing. You send it, from your own email, once you have checked it.
  2. “Does the program name any request or sample from these practice files? If it does, make it work those out from the files instead.” A program built in a conversation may keep the practice codes written into it, and next week’s requests would be checked wrongly.
  3. “Save a checkpoint.” Your AI tool saves a snapshot you can return to.

When you want to check real requests, use the finished intake check. It runs on your real files by itself, with your AI tool closed, so your AI tool never sees them. Real requests carry names, email addresses, account numbers and safety answers. Nothing confidential or unpublished, and nothing about people or about animals under a protocol, goes to your AI tool or into this course folder. Nobody hands you files shaped like the practice ones: your request system may export a spreadsheet with its own column names, or you may type each request in from the form. Typing them in is worth it only if it is quicker than checking by eye, and then the program checks what you typed; where your system exports, use the export. Do these steps in order; the first four happen now, while your AI tool is open and this folder still holds only the practice files:

  1. Go through the data check, the questions in setup about what the data is, who decides how it is used, and which AI tool would see it. Requests are your facility’s records and carry account numbers and details about labs’ samples; your facility director and whoever your facility names for data decide. Go on only if they allow this use.

  2. Decide on a folder of its own for your real files, outside this course folder, for example ~/intake-checks. Nothing has to be in it yet. If the folder is not there, ask your AI tool to make it: “Make an empty folder called intake-checks in my home folder.”

  3. Ask your AI tool for blank sheets to fill in, in that folder, for example:

    In ~/intake-checks, make blank requests and samples sheets my spreadsheet program can open, and a copy of service-requirements.csv. Show the program working there on one made-up request, then clear it.

    Check that it says the made-up request showed up on a list and that both sheets are blank again. The copy of service-requirements.csv still holds the practice facility’s made-up rules; you will replace them with your facility’s own. If your request system already exports a spreadsheet, tell your AI tool its column names and file names instead, and give it two rows you make up in the same shape, never a real row, so it can change the program to read files like that. On Windows, also ask: “How do I open ~/intake-checks in my spreadsheet program on Windows?”, and note the answer.

  4. Ask your AI tool for the exact line, for example: “Give me the exact one line to run the check myself on the sheets in ~/intake-checks, with full paths, saving the list there too.” Use your own folder’s name. If it offers to copy your real files into this folder instead, say no. Copy the line it gives you into a note. It will look something like this one, though yours may differ:

    Terminal window
    python3 /home/dana/bio-workbench/07-intake-check/check_intake.py /home/dana/intake-checks --out /home/dana/intake-checks/intake-report.html

    Read it left to right. python3 runs a Python program. Next comes the intake check, in this lesson’s folder; then the folder it reads the three files from; then --out and the list it writes there. /home/dana is the home folder of someone whose user name is dana; your own user name goes where dana is, and on a Mac the folders begin /Users/ instead of /home/. The line runs the program where it is, so leave this lesson’s folder in place. If you are unsure about any part, ask your AI tool now, while it is still open.

  5. Leave your AI tool the way you did at the start of this lesson.

  6. With your AI tool closed, open the blank sheets in your spreadsheet program and fill them in from the week’s requests, or save your system’s export into ~/intake-checks, and put your facility’s own requirements into the copy of service-requirements.csv. Save them as the same file type, in the same folder, with the same names: in your spreadsheet program, choose CSV in the save box, because it may otherwise save in its own format. Before you run the line, look at the sample names in the saved sheets: a spreadsheet may have turned one into a date or dropped a leading zero. Then paste the line from your note at the $ or % prompt and press Enter. The list is saved beside your real files, not in the course folder. To come back to the lesson folder later, type this lesson’s folder line again, the line near the top of this page that begins with mkdir.

Try another case

Before the quiz, write two or three sentences for yourself. The Ivanova lab calls: IVA-R6 is the only tube of that sample they have, 180 ng against your minimum of 250, and they ask whether they can bring it today anyway. What do you do, and what do you tell them?

A good answer leaves the requirements file alone and promises the lab nothing on the phone: not that the sample will be taken, and not that it will give usable data. It takes the question to whoever at the facility may make that call, under your own procedures, and writes the outcome, who decided and when, on R-1102. And it remembers the rest of R-1102: the missing approver, the duplicate IVA-R2 and the human-derived Yes still need answers before anything is cleared.

Key takeaways

  • A short ask is enough. Your job, the files and what you want; its questions fill in the rules.
  • The rules come from your facility’s written requirements and procedures, never from the tool’s suggestions.
  • A blank is missing, a different unit is a question, and a safety answer that needs a person goes to that person: none of them is guessed, converted or ruled on by the list.
  • The list is not the decision. Each request keeps a blank place for a person to clear it, ask the lab or hold it.
  • Practice data stays practice data: real requests go to the finished program, with your AI tool closed, after the data check.
KNOWLEDGE CHECK

R-1102's sample sheet lists IVA-R2 twice, with the same volume and concentration both times, and has 8 rows against 8 declared. What do you ask the lab?

KNOWLEDGE CHECK

MRS-03 has 450 ng against this practice facility's 500 ng minimum for GDNA-SEQ. Your manager agrees the Marsh lab may bring it as it is. What do you record, and where?

What’s next

The next lesson is about a core facility’s month-end charges: turning a month of instrument usage into a summary of each lab’s charges, checked before anything is billed.

intellimetrics · Arlington, VA
© 2026 intellimetrics.net
Search lessons